Extract v6 exit node pairing into shared helpers

# Conflicts:
#	client/firewall/uspfilter/forwarder/forwarder.go
#	client/firewall/uspfilter/forwarder/icmp.go

.devcontainer/Dockerfile

@@ -1,15 +1,15 @@
-FROM golang:1.23-bullseye
+FROM golang:1.25-bookworm
 
 RUN apt-get update && export DEBIAN_FRONTEND=noninteractive \
     && apt-get -y install --no-install-recommends\
-        gettext-base=0.21-4 \ 
-        iptables=1.8.7-1 \
-        libgl1-mesa-dev=20.3.5-1 \
-        xorg-dev=1:7.7+22 \
-        libayatana-appindicator3-dev=0.5.5-2+deb11u2 \
+        gettext-base=0.21-12 \
+        iptables=1.8.9-2 \
+        libgl1-mesa-dev=22.3.6-1+deb12u1 \
+        xorg-dev=1:7.7+23 \
+        libayatana-appindicator3-dev=0.5.92-1 \
     && apt-get clean \
     && rm -rf /var/lib/apt/lists/* \
-    && go install -v golang.org/x/tools/gopls@v0.18.1
+    && go install -v golang.org/x/tools/gopls@latest
 
 
 WORKDIR /app

.dockerignore

@@ -0,0 +1,6 @@
+.env
+.env.*
+*.pem
+*.key
+*.crt
+*.p12

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,14 @@
+blank_issues_enabled: true
+contact_links:
+  - name: Community Support
+    url: https://forum.netbird.io/
+    about: Community support forum
+  - name: Cloud Support
+    url: https://docs.netbird.io/help/report-bug-issues
+    about: Contact us for support
+  - name: Client/Connection Troubleshooting
+    url: https://docs.netbird.io/help/troubleshooting-client
+    about: See our client troubleshooting guide for help addressing common issues
+  - name: Self-host Troubleshooting
+    url: https://docs.netbird.io/selfhosted/troubleshooting
+    about: See our self-host troubleshooting guide for help addressing common issues

.github/workflows/check-license-dependencies.yml

@@ -23,7 +23,7 @@ jobs:
 
       - name: Check for problematic license dependencies
         run: |
-          echo "Checking for dependencies on management/, signal/, and relay/ packages..."
+          echo "Checking for dependencies on management/, signal/, relay/, and proxy/ packages..."
           echo ""
 
           # Find all directories except the problematic ones and system dirs
@@ -31,7 +31,7 @@ jobs:
           while IFS= read -r dir; do
             echo "=== Checking $dir ==="
             # Search for problematic imports, excluding test files
-            RESULTS=$(grep -r "github.com/netbirdio/netbird/\(management\|signal\|relay\)" "$dir" --include="*.go" 2>/dev/null | grep -v "_test.go" | grep -v "test_" | grep -v "/test/" || true)
+            RESULTS=$(grep -r "github.com/netbirdio/netbird/\(management\|signal\|relay\|proxy\)" "$dir" --include="*.go" 2>/dev/null | grep -v "_test.go" | grep -v "test_" | grep -v "/test/" || true)
             if [ -n "$RESULTS" ]; then
               echo "❌ Found problematic dependencies:"
               echo "$RESULTS"
@@ -39,11 +39,11 @@ jobs:
             else
               echo "✓ No problematic dependencies found"
             fi
-          done < <(find . -maxdepth 1 -type d -not -name "." -not -name "management" -not -name "signal" -not -name "relay" -not -name ".git*" | sort)
+          done < <(find . -maxdepth 1 -type d -not -name "." -not -name "management" -not -name "signal" -not -name "relay" -not -name "proxy" -not -name "combined" -not -name ".git*" | sort)
 
           echo ""
           if [ $FOUND_ISSUES -eq 1 ]; then
-            echo "❌ Found dependencies on management/, signal/, or relay/ packages"
+            echo "❌ Found dependencies on management/, signal/, relay/, or proxy/ packages"
             echo "These packages are licensed under AGPLv3 and must not be imported by BSD-licensed code"
             exit 1
           else
@@ -88,7 +88,7 @@ jobs:
               IMPORTERS=$(go list -json -deps ./... 2>/dev/null | jq -r "select(.Imports[]? == \"$package\") | .ImportPath")
 
               # Check if any importer is NOT in management/signal/relay
-              BSD_IMPORTER=$(echo "$IMPORTERS" | grep -v "github.com/netbirdio/netbird/\(management\|signal\|relay\)" | head -1)
+              BSD_IMPORTER=$(echo "$IMPORTERS" | grep -v "github.com/netbirdio/netbird/\(management\|signal\|relay\|proxy\|combined\)" | head -1)
 
               if [ -n "$BSD_IMPORTER" ]; then
                 echo "❌ $package ($license) is imported by BSD-licensed code: $BSD_IMPORTER"

.github/workflows/golang-test-darwin.yml

@@ -43,5 +43,5 @@ jobs:
         run: git --no-pager diff --exit-code
 
       - name: Test
-        run: NETBIRD_STORE_ENGINE=${{ matrix.store }} CI=true go test -tags=devcert -exec 'sudo --preserve-env=CI,NETBIRD_STORE_ENGINE' -timeout 5m -p 1 $(go list ./... | grep -v /management)
+        run: NETBIRD_STORE_ENGINE=${{ matrix.store }} CI=true go test -tags=devcert -exec 'sudo --preserve-env=CI,NETBIRD_STORE_ENGINE' -timeout 5m -p 1 $(go list ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined)
 

.github/workflows/golang-test-freebsd.yml

@@ -25,7 +25,7 @@ jobs:
           release: "14.2"
           prepare: |
             pkg install -y curl pkgconf xorg
-            GO_TARBALL="go1.24.10.freebsd-amd64.tar.gz"
+            GO_TARBALL="go1.25.3.freebsd-amd64.tar.gz"
             GO_URL="https://go.dev/dl/$GO_TARBALL"
             curl -vLO "$GO_URL"
             tar -C /usr/local -vxzf "$GO_TARBALL"            
@@ -46,6 +46,5 @@ jobs:
             time go test -timeout 1m -failfast ./client/iface/...
             time go test -timeout 1m -failfast ./route/...
             time go test -timeout 1m -failfast ./sharedsock/...
-            time go test -timeout 1m -failfast ./signal/...
             time go test -timeout 1m -failfast ./util/...
             time go test -timeout 1m -failfast ./version/...

.github/workflows/golang-test-linux.yml

@@ -97,6 +97,16 @@ jobs:
         working-directory: relay
         run: CGO_ENABLED=1 GOARCH=386 go build -o relay-386 .
 
+      - name: Build combined
+        if: steps.cache.outputs.cache-hit != 'true'
+        working-directory: combined
+        run: CGO_ENABLED=1 go build .
+
+      - name: Build combined 386
+        if: steps.cache.outputs.cache-hit != 'true'
+        working-directory: combined
+        run: CGO_ENABLED=1 GOARCH=386 go build -o combined-386 .
+
   test:
     name: "Client / Unit"
     needs: [build-cache]
@@ -144,7 +154,7 @@ jobs:
         run: git --no-pager diff --exit-code
 
       - name: Test
-        run: CGO_ENABLED=1 GOARCH=${{ matrix.arch }} CI=true go test -tags devcert -exec 'sudo' -timeout 10m -p 1 $(go list ./... | grep -v -e /management -e /signal -e /relay)
+        run: CGO_ENABLED=1 GOARCH=${{ matrix.arch }} CI=true go test -tags devcert -exec 'sudo' -timeout 10m -p 1 $(go list ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined)
 
   test_client_on_docker:
     name: "Client (Docker) / Unit"
@@ -200,11 +210,11 @@ jobs:
             -e GOCACHE=${CONTAINER_GOCACHE} \
             -e GOMODCACHE=${CONTAINER_GOMODCACHE} \
             -e CONTAINER=${CONTAINER} \
-            golang:1.24-alpine \
+            golang:1.25-alpine \
             sh -c ' \
               apk update; apk add --no-cache \
                 ca-certificates iptables ip6tables dbus dbus-dev libpcap-dev build-base; \
-              go test -buildvcs=false -tags devcert -v -timeout 10m -p 1 $(go list -buildvcs=false ./... | grep -v -e /management -e /signal -e /relay -e /client/ui -e /upload-server)
+              go test -buildvcs=false -tags devcert -v -timeout 10m -p 1 $(go list -buildvcs=false ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined -e /client/ui -e /upload-server)
             '
 
   test_relay:
@@ -259,7 +269,54 @@ jobs:
           CGO_ENABLED=1 GOARCH=${{ matrix.arch }} \
           go test ${{ matrix.raceFlag }} \
             -exec 'sudo' \
-            -timeout 10m ./relay/... ./shared/relay/...
+            -timeout 10m -p 1 ./relay/... ./shared/relay/...
+
+  test_proxy:
+    name: "Proxy / Unit"
+    needs: [build-cache]
+    strategy:
+      fail-fast: false
+      matrix:
+        arch: [ '386','amd64' ]
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Install Go
+        uses: actions/setup-go@v5
+        with:
+          go-version-file: "go.mod"
+          cache: false
+
+      - name: Install dependencies
+        run: sudo apt update && sudo apt install -y gcc-multilib g++-multilib libc6-dev-i386
+
+      - name: Get Go environment
+        run: |
+          echo "cache=$(go env GOCACHE)" >> $GITHUB_ENV
+          echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
+
+      - name: Cache Go modules
+        uses: actions/cache/restore@v4
+        with:
+          path: |
+            ${{ env.cache }}
+            ${{ env.modcache }}
+          key: ${{ runner.os }}-gotest-cache-${{ hashFiles('**/go.sum') }}
+          restore-keys: |
+            ${{ runner.os }}-gotest-cache-
+
+      - name: Install modules
+        run: go mod tidy
+
+      - name: check git status
+        run: git --no-pager diff --exit-code
+
+      - name: Test
+        run: |
+          CGO_ENABLED=1 GOARCH=${{ matrix.arch }} \
+          go test -timeout 10m -p 1 ./proxy/...
 
   test_signal:
     name: "Signal / Unit"
@@ -352,12 +409,19 @@ jobs:
         run: git --no-pager diff --exit-code
 
       - name: Login to Docker hub
-        if: matrix.store == 'mysql' && (github.repository == github.head.repo.full_name || !github.head_ref)
-        uses: docker/login-action@v1
+        if: github.event.pull_request && github.event.pull_request.head.repo && github.event.pull_request.head.repo.full_name == '' || github.repository == github.event.pull_request.head.repo.full_name || !github.head_ref
+        uses: docker/login-action@v3
         with:
           username: ${{ secrets.DOCKER_USER }}
           password: ${{ secrets.DOCKER_TOKEN }}
 
+      - name: docker login for root user
+        if: github.event.pull_request && github.event.pull_request.head.repo && github.event.pull_request.head.repo.full_name == '' || github.repository == github.event.pull_request.head.repo.full_name || !github.head_ref
+        env:
+          DOCKER_USER: ${{ secrets.DOCKER_USER }}
+          DOCKER_TOKEN: ${{ secrets.DOCKER_TOKEN }}
+        run: echo "$DOCKER_TOKEN" | sudo docker login --username "$DOCKER_USER" --password-stdin
+
       - name: download mysql image
         if: matrix.store == 'mysql'
         run: docker pull mlsmaycon/warmed-mysql:8
@@ -440,15 +504,18 @@ jobs:
         run: git --no-pager diff --exit-code
 
       - name: Login to Docker hub
-        if: matrix.store == 'mysql' && (github.repository == github.head.repo.full_name || !github.head_ref)
-        uses: docker/login-action@v1
+        if: github.event.pull_request && github.event.pull_request.head.repo && github.event.pull_request.head.repo.full_name == '' || github.repository == github.event.pull_request.head.repo.full_name || !github.head_ref
+        uses: docker/login-action@v3
         with:
           username: ${{ secrets.DOCKER_USER }}
           password: ${{ secrets.DOCKER_TOKEN }}
 
-      - name: download mysql image
-        if: matrix.store == 'mysql'
-        run: docker pull mlsmaycon/warmed-mysql:8
+      - name: docker login for root user
+        if: github.event.pull_request && github.event.pull_request.head.repo && github.event.pull_request.head.repo.full_name == '' || github.repository == github.event.pull_request.head.repo.full_name || !github.head_ref
+        env:
+          DOCKER_USER: ${{ secrets.DOCKER_USER }}
+          DOCKER_TOKEN: ${{ secrets.DOCKER_TOKEN }}
+        run: echo "$DOCKER_TOKEN" | sudo docker login --username "$DOCKER_USER" --password-stdin
 
       - name: Test
         run: |
@@ -529,15 +596,18 @@ jobs:
         run: git --no-pager diff --exit-code
 
       - name: Login to Docker hub
-        if: matrix.store == 'mysql' && (github.repository == github.head.repo.full_name || !github.head_ref)
-        uses: docker/login-action@v1
+        if: github.event.pull_request && github.event.pull_request.head.repo && github.event.pull_request.head.repo.full_name == '' || github.repository == github.event.pull_request.head.repo.full_name || !github.head_ref
+        uses: docker/login-action@v3
         with:
           username: ${{ secrets.DOCKER_USER }}
           password: ${{ secrets.DOCKER_TOKEN }}
 
-      - name: download mysql image
-        if: matrix.store == 'mysql'
-        run: docker pull mlsmaycon/warmed-mysql:8
+      - name: docker login for root user
+        if: github.event.pull_request && github.event.pull_request.head.repo && github.event.pull_request.head.repo.full_name == '' || github.repository == github.event.pull_request.head.repo.full_name || !github.head_ref
+        env:
+          DOCKER_USER: ${{ secrets.DOCKER_USER }}
+          DOCKER_TOKEN: ${{ secrets.DOCKER_TOKEN }}
+        run: echo "$DOCKER_TOKEN" | sudo docker login --username "$DOCKER_USER" --password-stdin
 
       - name: Test
         run: |

.github/workflows/golang-test-windows.yml

@@ -63,10 +63,15 @@ jobs:
       - run: PsExec64 -s -w ${{ github.workspace }} C:\hostedtoolcache\windows\go\${{ steps.go.outputs.go-version }}\x64\bin\go.exe env -w GOMODCACHE=${{ env.cache }}
       - run: PsExec64 -s -w ${{ github.workspace }} C:\hostedtoolcache\windows\go\${{ steps.go.outputs.go-version }}\x64\bin\go.exe env -w GOCACHE=${{ env.modcache }}
       - run: PsExec64 -s -w ${{ github.workspace }} C:\hostedtoolcache\windows\go\${{ steps.go.outputs.go-version }}\x64\bin\go.exe mod tidy
-      - run: echo "files=$(go list ./... | ForEach-Object { $_ } | Where-Object { $_ -notmatch '/management' } | Where-Object { $_ -notmatch '/relay' } | Where-Object { $_ -notmatch '/signal' })" >> $env:GITHUB_ENV
+      - name: Generate test script
+        run: |
+          $packages = go list ./... | Where-Object { $_ -notmatch '/management' } | Where-Object { $_ -notmatch '/relay' } | Where-Object { $_ -notmatch '/signal' } | Where-Object { $_ -notmatch '/proxy' } | Where-Object { $_ -notmatch '/combined' }
+          $goExe = "C:\hostedtoolcache\windows\go\${{ steps.go.outputs.go-version }}\x64\bin\go.exe"
+          $cmd = "$goExe test -tags=devcert -timeout 10m -p 1 $($packages -join ' ') > test-out.txt 2>&1"
+          Set-Content -Path "${{ github.workspace }}\run-tests.cmd" -Value $cmd
 
       - name: test
-        run: PsExec64 -s -w ${{ github.workspace }} cmd.exe /c "C:\hostedtoolcache\windows\go\${{ steps.go.outputs.go-version }}\x64\bin\go.exe test -tags=devcert -timeout 10m -p 1 ${{ env.files }} > test-out.txt 2>&1"
+        run: PsExec64 -s -w ${{ github.workspace }} cmd.exe /c "${{ github.workspace }}\run-tests.cmd"
       - name: test output
         if: ${{ always() }}
         run: Get-Content test-out.txt

.github/workflows/golangci-lint.yml

@@ -19,8 +19,8 @@ jobs:
       - name: codespell
         uses: codespell-project/actions-codespell@v2
         with:
-          ignore_words_list: erro,clienta,hastable,iif,groupd,testin,groupe,cros
-          skip: go.mod,go.sum
+          ignore_words_list: erro,clienta,hastable,iif,groupd,testin,groupe,cros,ans,deriver,te
+          skip: go.mod,go.sum,**/proxy/web/**
   golangci:
     strategy:
       fail-fast: false
@@ -52,7 +52,10 @@ jobs:
         if: matrix.os == 'ubuntu-latest'
         run: sudo apt update && sudo apt install -y -q libgtk-3-dev libayatana-appindicator3-dev libgl1-mesa-dev xorg-dev libpcap-dev
       - name: golangci-lint
-        uses: golangci/golangci-lint-action@v4
+        uses: golangci/golangci-lint-action@4afd733a84b1f43292c63897423277bb7f4313a9 # v8.0.0
         with:
           version: latest
-          args: --timeout=12m --out-format colored-line-number
+          skip-cache: true
+          skip-save-cache: true
+          cache-invalidation-interval: 0
+          args: --timeout=12m

.github/workflows/pr-title-check.yml

@@ -0,0 +1,51 @@
+name: PR Title Check
+
+on:
+  pull_request:
+    types: [opened, edited, synchronize, reopened]
+
+jobs:
+  check-title:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Validate PR title prefix
+        uses: actions/github-script@v7
+        with:
+          script: |
+            const title = context.payload.pull_request.title;
+            const allowedTags = [
+              'management',
+              'client',
+              'signal',
+              'proxy',
+              'relay',
+              'misc',
+              'infrastructure',
+              'self-hosted',
+              'doc',
+            ];
+
+            const pattern = /^\[([^\]]+)\]\s+.+/;
+            const match = title.match(pattern);
+
+            if (!match) {
+              core.setFailed(
+                `PR title must start with a tag in brackets.\n` +
+                `Example: [client] fix something\n` +
+                `Allowed tags: ${allowedTags.join(', ')}`
+              );
+              return;
+            }
+
+            const tags = match[1].split(',').map(t => t.trim().toLowerCase());
+
+            const invalid = tags.filter(t => !allowedTags.includes(t));
+            if (invalid.length > 0) {
+              core.setFailed(
+                `Invalid tag(s): ${invalid.join(', ')}\n` +
+                `Allowed tags: ${allowedTags.join(', ')}`
+              );
+              return;
+            }
+
+            console.log(`Valid PR title tags: [${tags.join(', ')}]`);

.github/workflows/release.yml

@@ -9,8 +9,8 @@ on:
   pull_request:
 
 env:
-  SIGN_PIPE_VER: "v0.1.0"
-  GORELEASER_VER: "v2.3.2"
+  SIGN_PIPE_VER: "v0.1.1"
+  GORELEASER_VER: "v2.14.3"
   PRODUCT_NAME: "NetBird"
   COPYRIGHT: "NetBird GmbH"
 
@@ -63,7 +63,7 @@ jobs:
             pkg install -y git curl portlint go
 
             # Install Go for building
-            GO_TARBALL="go1.24.10.freebsd-amd64.tar.gz"
+            GO_TARBALL="go1.25.5.freebsd-amd64.tar.gz"
             GO_URL="https://go.dev/dl/$GO_TARBALL"
             curl -LO "$GO_URL"
             tar -C /usr/local -xzf "$GO_TARBALL"
@@ -160,7 +160,7 @@ jobs:
           username: ${{ secrets.DOCKER_USER }}
           password: ${{ secrets.DOCKER_TOKEN }}
       - name: Log in to the GitHub container registry
-        if: github.event_name != 'pull_request'
+        if: github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository
         uses: docker/login-action@v3
         with:
           registry: ghcr.io
@@ -169,6 +169,14 @@ jobs:
       - name: Install OS build dependencies
         run: sudo apt update && sudo apt install -y -q gcc-arm-linux-gnueabihf gcc-aarch64-linux-gnu
 
+      - name: Decode GPG signing key
+        if: github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository
+        env:
+          GPG_RPM_PRIVATE_KEY: ${{ secrets.GPG_RPM_PRIVATE_KEY }}
+        run: |
+          echo "$GPG_RPM_PRIVATE_KEY" | base64 -d > /tmp/gpg-rpm-signing-key.asc
+          echo "GPG_RPM_KEY_FILE=/tmp/gpg-rpm-signing-key.asc" >> $GITHUB_ENV
+
       - name: Install goversioninfo
         run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@233067e
       - name: Generate windows syso amd64
@@ -176,6 +184,7 @@ jobs:
       - name: Generate windows syso arm64
         run: goversioninfo -arm -64 -icon client/ui/assets/netbird.ico -manifest client/manifest.xml -product-name ${{ env.PRODUCT_NAME }} -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/resources_windows_arm64.syso
       - name: Run GoReleaser
+        id: goreleaser
         uses: goreleaser/goreleaser-action@v4
         with:
           version: ${{ env.GORELEASER_VER }}
@@ -185,6 +194,55 @@ jobs:
           HOMEBREW_TAP_GITHUB_TOKEN: ${{ secrets.HOMEBREW_TAP_GITHUB_TOKEN }}
           UPLOAD_DEBIAN_SECRET: ${{ secrets.PKG_UPLOAD_SECRET }}
           UPLOAD_YUM_SECRET: ${{ secrets.PKG_UPLOAD_SECRET }}
+          GPG_RPM_KEY_FILE: ${{ env.GPG_RPM_KEY_FILE }}
+          NFPM_NETBIRD_RPM_PASSPHRASE: ${{ secrets.GPG_RPM_PASSPHRASE }}
+      - name: Verify RPM signatures
+        run: |
+          docker run --rm -v $(pwd)/dist:/dist fedora:41 bash -c '
+            dnf install -y -q rpm-sign curl >/dev/null 2>&1
+            curl -sSL https://pkgs.netbird.io/yum/repodata/repomd.xml.key -o /tmp/rpm-pub.key
+            rpm --import /tmp/rpm-pub.key
+            echo "=== Verifying RPM signatures ==="
+            for rpm_file in /dist/*amd64*.rpm; do
+              [ -f "$rpm_file" ] || continue
+              echo "--- $(basename $rpm_file) ---"
+              rpm -K "$rpm_file"
+            done
+          '
+      - name: Clean up GPG key
+        if: always()
+        run: rm -f /tmp/gpg-rpm-signing-key.asc
+      - name: Tag and push images (amd64 only)
+        if: |
+          (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository) ||
+          (github.event_name == 'push' && github.ref == 'refs/heads/main')
+        run: |
+          resolve_tags() {
+            if [[ "${{ github.event_name }}" == "pull_request" ]]; then
+              echo "pr-${{ github.event.pull_request.number }}"
+            else
+              echo "main sha-$(git rev-parse --short HEAD)"
+            fi
+          }
+
+          tag_and_push() {
+            local src="$1" img_name tag dst
+            img_name="${src%%:*}"
+            for tag in $(resolve_tags); do
+              dst="${img_name}:${tag}"
+              echo "Tagging ${src} -> ${dst}"
+              docker tag "$src" "$dst"
+              docker push "$dst"
+            done
+          }
+
+          export -f tag_and_push resolve_tags
+
+          echo '${{ steps.goreleaser.outputs.artifacts }}' | \
+            jq -r '.[] | select(.type == "Docker Image") | select(.goarch == "amd64") | .name' | \
+            grep '^ghcr.io/' | while read -r SRC; do
+              tag_and_push "$SRC"
+            done
       - name: upload non tags for debug purposes
         uses: actions/upload-artifact@v4
         with:
@@ -251,6 +309,14 @@ jobs:
       - name: Install dependencies
         run: sudo apt update && sudo apt install -y -q libappindicator3-dev gir1.2-appindicator3-0.1 libxxf86vm-dev gcc-mingw-w64-x86-64
 
+      - name: Decode GPG signing key
+        if: github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository
+        env:
+          GPG_RPM_PRIVATE_KEY: ${{ secrets.GPG_RPM_PRIVATE_KEY }}
+        run: |
+          echo "$GPG_RPM_PRIVATE_KEY" | base64 -d > /tmp/gpg-rpm-signing-key.asc
+          echo "GPG_RPM_KEY_FILE=/tmp/gpg-rpm-signing-key.asc" >> $GITHUB_ENV
+
       - name: Install LLVM-MinGW for ARM64 cross-compilation
         run: |
           cd /tmp
@@ -275,6 +341,24 @@ jobs:
           HOMEBREW_TAP_GITHUB_TOKEN: ${{ secrets.HOMEBREW_TAP_GITHUB_TOKEN }}
           UPLOAD_DEBIAN_SECRET: ${{ secrets.PKG_UPLOAD_SECRET }}
           UPLOAD_YUM_SECRET: ${{ secrets.PKG_UPLOAD_SECRET }}
+          GPG_RPM_KEY_FILE: ${{ env.GPG_RPM_KEY_FILE }}
+          NFPM_NETBIRD_UI_RPM_PASSPHRASE: ${{ secrets.GPG_RPM_PASSPHRASE }}
+      - name: Verify RPM signatures
+        run: |
+          docker run --rm -v $(pwd)/dist:/dist fedora:41 bash -c '
+            dnf install -y -q rpm-sign curl >/dev/null 2>&1
+            curl -sSL https://pkgs.netbird.io/yum/repodata/repomd.xml.key -o /tmp/rpm-pub.key
+            rpm --import /tmp/rpm-pub.key
+            echo "=== Verifying RPM signatures ==="
+            for rpm_file in /dist/*.rpm; do
+              [ -f "$rpm_file" ] || continue
+              echo "--- $(basename $rpm_file) ---"
+              rpm -K "$rpm_file"
+            done
+          '
+      - name: Clean up GPG key
+        if: always()
+        run: rm -f /tmp/gpg-rpm-signing-key.asc
       - name: upload non tags for debug purposes
         uses: actions/upload-artifact@v4
         with:

.github/workflows/wasm-build-validation.yml

@@ -14,6 +14,9 @@ jobs:
   js_lint:
     name: "JS / Lint"
     runs-on: ubuntu-latest
+    env:
+      GOOS: js
+      GOARCH: wasm
     steps:
       - name: Checkout repository
         uses: actions/checkout@v4
@@ -24,16 +27,14 @@ jobs:
       - name: Install dependencies
         run: sudo apt update && sudo apt install -y -q libgtk-3-dev libayatana-appindicator3-dev libgl1-mesa-dev xorg-dev libpcap-dev
       - name: Install golangci-lint
-        uses: golangci/golangci-lint-action@d6238b002a20823d52840fda27e2d4891c5952dc
+        uses: golangci/golangci-lint-action@4afd733a84b1f43292c63897423277bb7f4313a9 # v8.0.0
         with:
           version: latest
           install-mode: binary
           skip-cache: true
-          skip-pkg-cache: true
-          skip-build-cache: true
-      - name: Run golangci-lint for WASM
-        run: |
-          GOOS=js GOARCH=wasm golangci-lint run --timeout=12m --out-format colored-line-number ./client/...
+          skip-save-cache: true
+          cache-invalidation-interval: 0
+          working-directory: ./client
         continue-on-error: true
 
   js_build:
@@ -60,8 +61,8 @@ jobs:
 
           echo "Size: ${SIZE} bytes (${SIZE_MB} MB)"
 
-          if [ ${SIZE} -gt 57671680 ]; then
-            echo "Wasm binary size (${SIZE_MB}MB) exceeds 55MB limit!"
+          if [ ${SIZE} -gt 58720256 ]; then
+            echo "Wasm binary size (${SIZE_MB}MB) exceeds 56MB limit!"
             exit 1
           fi
 

.gitignore

@@ -2,6 +2,7 @@
 .run
 *.iml
 dist/
+!proxy/web/dist/
 bin/
 .env
 conf.json

.golangci.yaml

@@ -1,139 +1,124 @@
-run:
-  # Timeout for analysis, e.g. 30s, 5m.
-  # Default: 1m
-  timeout: 6m
-
-# This file contains only configs which differ from defaults.
-# All possible options can be found here https://github.com/golangci/golangci-lint/blob/master/.golangci.reference.yml
-linters-settings:
-  errcheck:
-    # Report about not checking of errors in type assertions: `a := b.(MyStruct)`.
-    # Such cases aren't reported by default.
-    # Default: false
-    check-type-assertions: false
-
-  gosec:
-    includes:
-      - G101 # Look for hard coded credentials
-      #- G102 # Bind to all interfaces
-      - G103 # Audit the use of unsafe block
-      - G104 # Audit errors not checked
-      - G106 # Audit the use of ssh.InsecureIgnoreHostKey
-      #- G107 # Url provided to HTTP request as taint input
-      - G108 # Profiling endpoint automatically exposed on /debug/pprof
-      - G109 # Potential Integer overflow made by strconv.Atoi result conversion to int16/32
-      - G110 # Potential DoS vulnerability via decompression bomb
-      - G111 # Potential directory traversal
-      #- G112 # Potential slowloris attack
-      - G113 # Usage of Rat.SetString in math/big with an overflow (CVE-2022-23772)
-      #- G114 # Use of net/http serve function that has no support for setting timeouts
-      - G201 # SQL query construction using format string
-      - G202 # SQL query construction using string concatenation
-      - G203 # Use of unescaped data in HTML templates
-      #- G204 # Audit use of command execution
-      - G301 # Poor file permissions used when creating a directory
-      - G302 # Poor file permissions used with chmod
-      - G303 # Creating tempfile using a predictable path
-      - G304 # File path provided as taint input
-      - G305 # File traversal when extracting zip/tar archive
-      - G306 # Poor file permissions used when writing to a new file
-      - G307 # Poor file permissions used when creating a file with os.Create
-      #- G401 # Detect the usage of DES, RC4, MD5 or SHA1
-      #- G402 # Look for bad TLS connection settings
-      - G403 # Ensure minimum RSA key length of 2048 bits
-      #- G404 # Insecure random number source (rand)
-      #- G501 # Import blocklist: crypto/md5
-      - G502 # Import blocklist: crypto/des
-      - G503 # Import blocklist: crypto/rc4
-      - G504 # Import blocklist: net/http/cgi
-      #- G505 # Import blocklist: crypto/sha1
-      - G601 # Implicit memory aliasing of items from a range statement
-      - G602 # Slice access out of bounds
-
-  gocritic:
-    disabled-checks:
-      - commentFormatting
-      - captLocal
-      - deprecatedComment
-
-  govet:
-    # Enable all analyzers.
-    # Default: false
-    enable-all: false
-    enable:
-      - nilness
-
-  revive:
-    rules:
-      - name: exported
-        severity: warning
-        disabled: false
-        arguments:
-          - "checkPrivateReceivers"
-          - "sayRepetitiveInsteadOfStutters"
-  tenv:
-    # The option `all` will run against whole test files (`_test.go`) regardless of method/function signatures.
-    # Otherwise, only methods that take `*testing.T`, `*testing.B`, and `testing.TB` as arguments are checked.
-    # Default: false
-    all: true
-
+version: "2"
 linters:
-  disable-all: true
+  default: none
   enable:
-    ## enabled by default
-    - errcheck # checking for unchecked errors, these unchecked errors can be critical bugs in some cases
-    - gosimple # specializes in simplifying a code
-    - govet # reports suspicious constructs, such as Printf calls whose arguments do not align with the format string
-    - ineffassign # detects when assignments to existing variables are not used
-    - staticcheck # is a go vet on steroids, applying a ton of static analysis checks
-    - tenv # Tenv is analyzer that detects using os.Setenv instead of t.Setenv since Go1.17.
-    - typecheck # like the front-end of a Go compiler, parses and type-checks Go code
-    - unused # checks for unused constants, variables, functions and types
-    ## disable by default but the have interesting results so lets add them
-    - bodyclose # checks whether HTTP response body is closed successfully
-    - dupword # dupword checks for duplicate words in the source code
-    - durationcheck # durationcheck checks for two durations multiplied together
-    - forbidigo # forbidigo forbids identifiers
-    - gocritic # provides diagnostics that check for bugs, performance and style issues
-    - gosec # inspects source code for security problems
-    - mirror # mirror reports wrong mirror patterns of bytes/strings usage
-    - misspell # misspess finds commonly misspelled English words in comments
-    - nilerr # finds the code that returns nil even if it checks that the error is not nil
-    - nilnil # checks that there is no simultaneous return of nil error and an invalid value
-    - predeclared # predeclared finds code that shadows one of Go's predeclared identifiers
-    - revive # Fast, configurable, extensible, flexible, and beautiful linter for Go. Drop-in replacement of golint.
-    - sqlclosecheck # checks that sql.Rows and sql.Stmt are closed
-    # - thelper # thelper detects Go test helpers without t.Helper() call and checks the consistency of test helpers.
-    - wastedassign # wastedassign finds wasted assignment statements
+    - bodyclose
+    - dupword
+    - durationcheck
+    - errcheck
+    - forbidigo
+    - gocritic
+    - gosec
+    - govet
+    - ineffassign
+    - mirror
+    - misspell
+    - nilerr
+    - nilnil
+    - predeclared
+    - revive
+    - sqlclosecheck
+    - staticcheck
+    - unused
+    - wastedassign
+  settings:
+    errcheck:
+      check-type-assertions: false
+    gocritic:
+      disabled-checks:
+        - commentFormatting
+        - captLocal
+        - deprecatedComment
+    gosec:
+      includes:
+        - G101
+        - G103
+        - G104
+        - G106
+        - G108
+        - G109
+        - G110
+        - G111
+        - G201
+        - G202
+        - G203
+        - G301
+        - G302
+        - G303
+        - G304
+        - G305
+        - G306
+        - G307
+        - G403
+        - G502
+        - G503
+        - G504
+        - G601
+        - G602
+    govet:
+      enable:
+        - nilness
+      enable-all: false
+    revive:
+      rules:
+        - name: exported
+          arguments:
+            - checkPrivateReceivers
+            - sayRepetitiveInsteadOfStutters
+          severity: warning
+          disabled: false
+  exclusions:
+    generated: lax
+    presets:
+      - comments
+      - common-false-positives
+      - legacy
+      - std-error-handling
+    rules:
+      - linters:
+          - forbidigo
+        path: management/cmd/root\.go
+      - linters:
+          - forbidigo
+        path: signal/cmd/root\.go
+      - linters:
+          - unused
+        path: sharedsock/filter\.go
+      - linters:
+          - unused
+        path: client/firewall/iptables/rule\.go
+      - linters:
+          - gosec
+          - mirror
+        path: test\.go
+      - linters:
+          - nilnil
+        path: mock\.go
+      - linters:
+          - staticcheck
+        text: grpc.DialContext is deprecated
+      - linters:
+          - staticcheck
+        text: grpc.WithBlock is deprecated
+      - linters:
+          - staticcheck
+        text: "QF1001"
+      - linters:
+          - staticcheck
+        text: "QF1008"
+      - linters:
+          - staticcheck
+        text: "QF1012"
+    paths:
+      - third_party$
+      - builtin$
+      - examples$
 issues:
-  # Maximum count of issues with the same text.
-  # Set to 0 to disable.
-  # Default: 3
   max-same-issues: 5
-
-  exclude-rules:
-    # allow fmt
-    - path: management/cmd/root\.go
-      linters: forbidigo
-    - path: signal/cmd/root\.go
-      linters: forbidigo
-    - path: sharedsock/filter\.go
-      linters:
-      - unused
-    - path: client/firewall/iptables/rule\.go
-      linters:
-      - unused
-    - path: test\.go
-      linters:
-      - mirror
-      - gosec
-    - path: mock\.go
-      linters:
-      - nilnil
-    # Exclude specific deprecation warnings for grpc methods
-    - linters:
-       - staticcheck
-      text: "grpc.DialContext is deprecated"
-    - linters:
-        - staticcheck
-      text: "grpc.WithBlock is deprecated"
+formatters:
+  exclusions:
+    generated: lax
+    paths:
+      - third_party$
+      - builtin$
+      - examples$

.goreleaser.yaml

@@ -106,6 +106,26 @@ builds:
       - -s -w -X github.com/netbirdio/netbird/version.version={{.Version}} -X main.commit={{.Commit}} -X main.date={{.CommitDate}} -X main.builtBy=goreleaser
     mod_timestamp: "{{ .CommitTimestamp }}"
 
+  - id: netbird-server
+    dir: combined
+    env:
+      - CGO_ENABLED=1
+      - >-
+        {{- if eq .Runtime.Goos "linux" }}
+          {{- if eq .Arch "arm64"}}CC=aarch64-linux-gnu-gcc{{- end }}
+          {{- if eq .Arch "arm"}}CC=arm-linux-gnueabihf-gcc{{- end }}
+        {{- end }}
+    binary: netbird-server
+    goos:
+      - linux
+    goarch:
+      - amd64
+      - arm64
+      - arm
+    ldflags:
+      - -s -w -X github.com/netbirdio/netbird/version.version={{.Version}} -X main.commit={{.Commit}} -X main.date={{.CommitDate}} -X main.builtBy=goreleaser
+    mod_timestamp: "{{ .CommitTimestamp }}"
+
   - id: netbird-upload
     dir: upload-server
     env: [CGO_ENABLED=0]
@@ -120,6 +140,20 @@ builds:
       - -s -w -X github.com/netbirdio/netbird/version.version={{.Version}} -X main.commit={{.Commit}} -X main.date={{.CommitDate}} -X main.builtBy=goreleaser
     mod_timestamp: "{{ .CommitTimestamp }}"
 
+  - id: netbird-proxy
+    dir: proxy/cmd/proxy
+    env: [CGO_ENABLED=0]
+    binary: netbird-proxy
+    goos:
+      - linux
+    goarch:
+      - amd64
+      - arm64
+      - arm
+    ldflags:
+      - -s -w -X main.Version={{.Version}} -X main.Commit={{.Commit}} -X main.BuildDate={{.CommitDate}}
+    mod_timestamp: "{{ .CommitTimestamp }}"
+
 universal_binaries:
   - id: netbird
 
@@ -137,13 +171,13 @@ nfpms:
   - maintainer: Netbird <dev@netbird.io>
     description: Netbird client.
     homepage: https://netbird.io/
-    id: netbird-deb
+    license: BSD-3-Clause
+    id: netbird_deb
     bindir: /usr/bin
     builds:
       - netbird
     formats:
       - deb
-
     scripts:
       postinstall: "release_files/post_install.sh"
       preremove: "release_files/pre_remove.sh"
@@ -151,16 +185,19 @@ nfpms:
   - maintainer: Netbird <dev@netbird.io>
     description: Netbird client.
     homepage: https://netbird.io/
-    id: netbird-rpm
+    license: BSD-3-Clause
+    id: netbird_rpm
     bindir: /usr/bin
     builds:
       - netbird
     formats:
       - rpm
-
     scripts:
       postinstall: "release_files/post_install.sh"
       preremove: "release_files/pre_remove.sh"
+    rpm:
+      signature:
+        key_file: '{{ if index .Env "GPG_RPM_KEY_FILE" }}{{ .Env.GPG_RPM_KEY_FILE }}{{ end }}'
 dockers:
   - image_templates:
       - netbirdio/netbird:{{ .Version }}-amd64
@@ -520,6 +557,104 @@ dockers:
       - "--label=org.opencontainers.image.revision={{.FullCommit}}"
       - "--label=org.opencontainers.image.source=https://github.com/netbirdio/{{.ProjectName}}"
       - "--label=maintainer=dev@netbird.io"
+  - image_templates:
+      - netbirdio/netbird-server:{{ .Version }}-amd64
+      - ghcr.io/netbirdio/netbird-server:{{ .Version }}-amd64
+    ids:
+      - netbird-server
+    goarch: amd64
+    use: buildx
+    dockerfile: combined/Dockerfile
+    build_flag_templates:
+      - "--platform=linux/amd64"
+      - "--label=org.opencontainers.image.created={{.Date}}"
+      - "--label=org.opencontainers.image.title={{.ProjectName}}"
+      - "--label=org.opencontainers.image.version={{.Version}}"
+      - "--label=org.opencontainers.image.revision={{.FullCommit}}"
+      - "--label=org.opencontainers.image.source=https://github.com/netbirdio/{{.ProjectName}}"
+      - "--label=maintainer=dev@netbird.io"
+  - image_templates:
+      - netbirdio/netbird-server:{{ .Version }}-arm64v8
+      - ghcr.io/netbirdio/netbird-server:{{ .Version }}-arm64v8
+    ids:
+      - netbird-server
+    goarch: arm64
+    use: buildx
+    dockerfile: combined/Dockerfile
+    build_flag_templates:
+      - "--platform=linux/arm64"
+      - "--label=org.opencontainers.image.created={{.Date}}"
+      - "--label=org.opencontainers.image.title={{.ProjectName}}"
+      - "--label=org.opencontainers.image.version={{.Version}}"
+      - "--label=org.opencontainers.image.revision={{.FullCommit}}"
+      - "--label=org.opencontainers.image.source=https://github.com/netbirdio/{{.ProjectName}}"
+      - "--label=maintainer=dev@netbird.io"
+  - image_templates:
+      - netbirdio/netbird-server:{{ .Version }}-arm
+      - ghcr.io/netbirdio/netbird-server:{{ .Version }}-arm
+    ids:
+      - netbird-server
+    goarch: arm
+    goarm: 6
+    use: buildx
+    dockerfile: combined/Dockerfile
+    build_flag_templates:
+      - "--platform=linux/arm"
+      - "--label=org.opencontainers.image.created={{.Date}}"
+      - "--label=org.opencontainers.image.title={{.ProjectName}}"
+      - "--label=org.opencontainers.image.version={{.Version}}"
+      - "--label=org.opencontainers.image.revision={{.FullCommit}}"
+      - "--label=org.opencontainers.image.source=https://github.com/netbirdio/{{.ProjectName}}"
+      - "--label=maintainer=dev@netbird.io"
+  - image_templates:
+      - netbirdio/reverse-proxy:{{ .Version }}-amd64
+      - ghcr.io/netbirdio/reverse-proxy:{{ .Version }}-amd64
+    ids:
+      - netbird-proxy
+    goarch: amd64
+    use: buildx
+    dockerfile: proxy/Dockerfile
+    build_flag_templates:
+      - "--platform=linux/amd64"
+      - "--label=org.opencontainers.image.created={{.Date}}"
+      - "--label=org.opencontainers.image.title={{.ProjectName}}"
+      - "--label=org.opencontainers.image.version={{.Version}}"
+      - "--label=org.opencontainers.image.revision={{.FullCommit}}"
+      - "--label=org.opencontainers.image.source=https://github.com/netbirdio/{{.ProjectName}}"
+      - "--label=maintainer=dev@netbird.io"
+  - image_templates:
+      - netbirdio/reverse-proxy:{{ .Version }}-arm64v8
+      - ghcr.io/netbirdio/reverse-proxy:{{ .Version }}-arm64v8
+    ids:
+      - netbird-proxy
+    goarch: arm64
+    use: buildx
+    dockerfile: proxy/Dockerfile
+    build_flag_templates:
+      - "--platform=linux/arm64"
+      - "--label=org.opencontainers.image.created={{.Date}}"
+      - "--label=org.opencontainers.image.title={{.ProjectName}}"
+      - "--label=org.opencontainers.image.version={{.Version}}"
+      - "--label=org.opencontainers.image.revision={{.FullCommit}}"
+      - "--label=org.opencontainers.image.source=https://github.com/netbirdio/{{.ProjectName}}"
+      - "--label=maintainer=dev@netbird.io"
+  - image_templates:
+      - netbirdio/reverse-proxy:{{ .Version }}-arm
+      - ghcr.io/netbirdio/reverse-proxy:{{ .Version }}-arm
+    ids:
+      - netbird-proxy
+    goarch: arm
+    goarm: 6
+    use: buildx
+    dockerfile: proxy/Dockerfile
+    build_flag_templates:
+      - "--platform=linux/arm"
+      - "--label=org.opencontainers.image.created={{.Date}}"
+      - "--label=org.opencontainers.image.title={{.ProjectName}}"
+      - "--label=org.opencontainers.image.version={{.Version}}"
+      - "--label=org.opencontainers.image.revision={{.FullCommit}}"
+      - "--label=org.opencontainers.image.source=https://github.com/netbirdio/{{.ProjectName}}"
+      - "--label=maintainer=dev@netbird.io"
 docker_manifests:
   - name_template: netbirdio/netbird:{{ .Version }}
     image_templates:
@@ -598,6 +733,18 @@ docker_manifests:
       - netbirdio/upload:{{ .Version }}-arm
       - netbirdio/upload:{{ .Version }}-amd64
 
+  - name_template: netbirdio/netbird-server:{{ .Version }}
+    image_templates:
+      - netbirdio/netbird-server:{{ .Version }}-arm64v8
+      - netbirdio/netbird-server:{{ .Version }}-arm
+      - netbirdio/netbird-server:{{ .Version }}-amd64
+
+  - name_template: netbirdio/netbird-server:latest
+    image_templates:
+      - netbirdio/netbird-server:{{ .Version }}-arm64v8
+      - netbirdio/netbird-server:{{ .Version }}-arm
+      - netbirdio/netbird-server:{{ .Version }}-amd64
+
   - name_template: ghcr.io/netbirdio/netbird:{{ .Version }}
     image_templates:
       - ghcr.io/netbirdio/netbird:{{ .Version }}-arm64v8
@@ -675,6 +822,43 @@ docker_manifests:
       - ghcr.io/netbirdio/upload:{{ .Version }}-arm64v8
       - ghcr.io/netbirdio/upload:{{ .Version }}-arm
       - ghcr.io/netbirdio/upload:{{ .Version }}-amd64
+
+  - name_template: ghcr.io/netbirdio/netbird-server:{{ .Version }}
+    image_templates:
+      - ghcr.io/netbirdio/netbird-server:{{ .Version }}-arm64v8
+      - ghcr.io/netbirdio/netbird-server:{{ .Version }}-arm
+      - ghcr.io/netbirdio/netbird-server:{{ .Version }}-amd64
+
+  - name_template: ghcr.io/netbirdio/netbird-server:latest
+    image_templates:
+      - ghcr.io/netbirdio/netbird-server:{{ .Version }}-arm64v8
+      - ghcr.io/netbirdio/netbird-server:{{ .Version }}-arm
+      - ghcr.io/netbirdio/netbird-server:{{ .Version }}-amd64
+
+  - name_template: netbirdio/reverse-proxy:{{ .Version }}
+    image_templates:
+      - netbirdio/reverse-proxy:{{ .Version }}-arm64v8
+      - netbirdio/reverse-proxy:{{ .Version }}-arm
+      - netbirdio/reverse-proxy:{{ .Version }}-amd64
+
+  - name_template: netbirdio/reverse-proxy:latest
+    image_templates:
+      - netbirdio/reverse-proxy:{{ .Version }}-arm64v8
+      - netbirdio/reverse-proxy:{{ .Version }}-arm
+      - netbirdio/reverse-proxy:{{ .Version }}-amd64
+
+  - name_template: ghcr.io/netbirdio/reverse-proxy:{{ .Version }}
+    image_templates:
+      - ghcr.io/netbirdio/reverse-proxy:{{ .Version }}-arm64v8
+      - ghcr.io/netbirdio/reverse-proxy:{{ .Version }}-arm
+      - ghcr.io/netbirdio/reverse-proxy:{{ .Version }}-amd64
+
+  - name_template: ghcr.io/netbirdio/reverse-proxy:latest
+    image_templates:
+      - ghcr.io/netbirdio/reverse-proxy:{{ .Version }}-arm64v8
+      - ghcr.io/netbirdio/reverse-proxy:{{ .Version }}-arm
+      - ghcr.io/netbirdio/reverse-proxy:{{ .Version }}-amd64
+
 brews:
   - ids:
       - default
@@ -695,7 +879,7 @@ brews:
 uploads:
   - name: debian
     ids:
-      - netbird-deb
+      - netbird_deb
     mode: archive
     target: https://pkgs.wiretrustee.com/debian/pool/{{ .ArtifactName }};deb.distribution=stable;deb.component=main;deb.architecture={{ if .Arm }}armhf{{ else }}{{ .Arch }}{{ end }};deb.package=
     username: dev@wiretrustee.com
@@ -703,7 +887,7 @@ uploads:
 
   - name: yum
     ids:
-      - netbird-rpm
+      - netbird_rpm
     mode: archive
     target: https://pkgs.wiretrustee.com/yum/{{ .Arch }}{{ if .Arm }}{{ .Arm }}{{ end }}
     username: dev@wiretrustee.com

.goreleaser_ui.yaml

@@ -61,7 +61,7 @@ nfpms:
   - maintainer: Netbird <dev@netbird.io>
     description: Netbird client UI.
     homepage: https://netbird.io/
-    id: netbird-ui-deb
+    id: netbird_ui_deb
     package_name: netbird-ui
     builds:
       - netbird-ui
@@ -80,7 +80,7 @@ nfpms:
   - maintainer: Netbird <dev@netbird.io>
     description: Netbird client UI.
     homepage: https://netbird.io/
-    id: netbird-ui-rpm
+    id: netbird_ui_rpm
     package_name: netbird-ui
     builds:
       - netbird-ui
@@ -95,11 +95,14 @@ nfpms:
         dst: /usr/share/pixmaps/netbird.png
     dependencies:
       - netbird
+    rpm:
+      signature:
+        key_file: '{{ if index .Env "GPG_RPM_KEY_FILE" }}{{ .Env.GPG_RPM_KEY_FILE }}{{ end }}'
 
 uploads:
   - name: debian
     ids:
-      - netbird-ui-deb
+      - netbird_ui_deb
     mode: archive
     target: https://pkgs.wiretrustee.com/debian/pool/{{ .ArtifactName }};deb.distribution=stable;deb.component=main;deb.architecture={{ if .Arm }}armhf{{ else }}{{ .Arch }}{{ end }};deb.package=
     username: dev@wiretrustee.com
@@ -107,7 +110,7 @@ uploads:
 
   - name: yum
     ids:
-      - netbird-ui-rpm
+      - netbird_ui_rpm
     mode: archive
     target: https://pkgs.wiretrustee.com/yum/{{ .Arch }}{{ if .Arm }}{{ .Arm }}{{ end }}
     username: dev@wiretrustee.com

LICENSE

@@ -1,4 +1,4 @@
-This BSD‑3‑Clause license applies to all parts of the repository except for the directories management/, signal/ and relay/.
+This BSD‑3‑Clause license applies to all parts of the repository except for the directories management/, signal/, relay/ and combined/.
 Those directories are licensed under the GNU Affero General Public License version 3.0 (AGPLv3). See the respective LICENSE files inside each directory.
 
 BSD 3-Clause License

README.md

@@ -38,6 +38,11 @@
  
 </strong>
 <br>
+<strong>
+  🚀 <a href="https://careers.netbird.io">We are hiring! Join us at careers.netbird.io</a>
+</strong>
+<br>
+<br>
 <a href="https://registry.terraform.io/providers/netbirdio/netbird/latest">
     New: NetBird terraform provider
   </a> 
@@ -55,8 +60,8 @@
 
 https://github.com/user-attachments/assets/10cec749-bb56-4ab3-97af-4e38850108d2
 
-### NetBird on Lawrence Systems (Video)
-[![Watch the video](https://img.youtube.com/vi/Kwrff6h0rEw/0.jpg)](https://www.youtube.com/watch?v=Kwrff6h0rEw)
+### Self-Host NetBird (Video)
+[![Watch the video](https://img.youtube.com/vi/bZAgpT6nzaQ/0.jpg)](https://youtu.be/bZAgpT6nzaQ)
 
 ### Key features
 
@@ -121,6 +126,7 @@ See a complete [architecture overview](https://docs.netbird.io/about-netbird/how
 ### Community projects
 -  [NetBird installer script](https://github.com/physk/netbird-installer)
 -  [NetBird ansible collection by Dominion Solutions](https://galaxy.ansible.com/ui/repo/published/dominion_solutions/netbird/)
+-  [netbird-tui](https://github.com/n0pashkov/netbird-tui) — terminal UI for managing NetBird peers, routes, and settings
 
 **Note**: The `main` branch may be in an *unstable or even broken state* during development.
 For stable versions, see [releases](https://github.com/netbirdio/netbird/releases).

client/Dockerfile

@@ -4,7 +4,7 @@
 #   sudo podman build -t localhost/netbird:latest -f client/Dockerfile --ignorefile .dockerignore-client .
 #   sudo podman run --rm -it --cap-add={BPF,NET_ADMIN,NET_RAW} localhost/netbird:latest
 
-FROM alpine:3.22.2
+FROM alpine:3.23.3
 # iproute2: busybox doesn't display ip rules properly
 RUN apk add --no-cache \
     bash \
@@ -17,8 +17,7 @@ ENV \
     NETBIRD_BIN="/usr/local/bin/netbird" \
     NB_LOG_FILE="console,/var/log/netbird/client.log" \
     NB_DAEMON_ADDR="unix:///var/run/netbird.sock" \
-    NB_ENTRYPOINT_SERVICE_TIMEOUT="5" \
-    NB_ENTRYPOINT_LOGIN_TIMEOUT="5"
+    NB_ENTRYPOINT_SERVICE_TIMEOUT="30"
 
 ENTRYPOINT [ "/usr/local/bin/netbird-entrypoint.sh" ]
 

client/Dockerfile-rootless

@@ -23,8 +23,7 @@ ENV \
     NB_DAEMON_ADDR="unix:///var/lib/netbird/netbird.sock" \
     NB_LOG_FILE="console,/var/lib/netbird/client.log" \
     NB_DISABLE_DNS="true" \
-    NB_ENTRYPOINT_SERVICE_TIMEOUT="5" \
-    NB_ENTRYPOINT_LOGIN_TIMEOUT="1"
+    NB_ENTRYPOINT_SERVICE_TIMEOUT="30"
 
 ENTRYPOINT [ "/usr/local/bin/netbird-entrypoint.sh" ]
 

client/android/client.go

@@ -124,7 +124,7 @@ func (c *Client) Run(platformFiles PlatformFiles, urlOpener URLOpener, isAndroid
 
 	// todo do not throw error in case of cancelled context
 	ctx = internal.CtxInitState(ctx)
-	c.connectClient = internal.NewConnectClient(ctx, cfg, c.recorder, false)
+	c.connectClient = internal.NewConnectClient(ctx, cfg, c.recorder)
 	return c.connectClient.RunOnAndroid(c.tunAdapter, c.iFaceDiscover, c.networkChangeListener, slices.Clone(dns.items), dnsReadyListener, stateFile)
 }
 
@@ -157,7 +157,7 @@ func (c *Client) RunWithoutLogin(platformFiles PlatformFiles, dns *DNSList, dnsR
 
 	// todo do not throw error in case of cancelled context
 	ctx = internal.CtxInitState(ctx)
-	c.connectClient = internal.NewConnectClient(ctx, cfg, c.recorder, false)
+	c.connectClient = internal.NewConnectClient(ctx, cfg, c.recorder)
 	return c.connectClient.RunOnAndroid(c.tunAdapter, c.iFaceDiscover, c.networkChangeListener, slices.Clone(dns.items), dnsReadyListener, stateFile)
 }
 
@@ -203,10 +203,11 @@ func (c *Client) PeersList() *PeerInfoArray {
 	peerInfos := make([]PeerInfo, len(fullStatus.Peers))
 	for n, p := range fullStatus.Peers {
 		pi := PeerInfo{
-			p.IP,
-			p.FQDN,
-			p.ConnStatus.String(),
-			PeerRoutes{routes: maps.Keys(p.GetRoutes())},
+			IP:         p.IP,
+			IPv6:       p.IPv6,
+			FQDN:       p.FQDN,
+			ConnStatus: p.ConnStatus.String(),
+			Routes:     PeerRoutes{routes: maps.Keys(p.GetRoutes())},
 		}
 		peerInfos[n] = pi
 	}
@@ -237,43 +238,84 @@ func (c *Client) Networks() *NetworkArray {
 		return nil
 	}
 
+	routesMap := routeManager.GetClientRoutesWithNetID()
+	v6Merged := route.V6ExitMergeSet(routesMap)
+	resolvedDomains := c.recorder.GetResolvedDomainsStates()
+
 	networkArray := &NetworkArray{
 		items: make([]Network, 0),
 	}
 
-	resolvedDomains := c.recorder.GetResolvedDomainsStates()
-
-	for id, routes := range routeManager.GetClientRoutesWithNetID() {
+	for id, routes := range routesMap {
 		if len(routes) == 0 {
 			continue
 		}
-
-		r := routes[0]
-		domains := c.getNetworkDomainsFromRoute(r, resolvedDomains)
-		netStr := r.Network.String()
-
-		if r.IsDynamic() {
-			netStr = r.Domains.SafeString()
-		}
-
-		routePeer, err := c.recorder.GetPeer(routes[0].Peer)
-		if err != nil {
-			log.Errorf("could not get peer info for %s: %v", routes[0].Peer, err)
+		if _, skip := v6Merged[id]; skip {
 			continue
 		}
-		network := Network{
-			Name:       string(id),
-			Network:    netStr,
-			Peer:       routePeer.FQDN,
-			Status:     routePeer.ConnStatus.String(),
-			IsSelected: routeSelector.IsSelected(id),
-			Domains:    domains,
+
+		network := c.buildNetwork(id, routes, routeSelector.IsSelected(id), resolvedDomains, v6Merged)
+		if network == nil {
+			continue
 		}
-		networkArray.Add(network)
+		networkArray.Add(*network)
 	}
 	return networkArray
 }
 
+func (c *Client) buildNetwork(id route.NetID, routes []*route.Route, selected bool, resolvedDomains map[domain.Domain]peer.ResolvedDomainInfo, v6Merged map[route.NetID]struct{}) *Network {
+	r := routes[0]
+	netStr := r.Network.String()
+	if r.IsDynamic() {
+		netStr = r.Domains.SafeString()
+	}
+
+	routePeer, err := c.findBestRoutePeer(routes)
+	if err != nil {
+		log.Errorf("could not get peer info for route %s: %v", id, err)
+		return nil
+	}
+
+	network := &Network{
+		Name:       string(id),
+		Network:    netStr,
+		Peer:       routePeer.FQDN,
+		Status:     routePeer.ConnStatus.String(),
+		IsSelected: selected,
+		Domains:    c.getNetworkDomainsFromRoute(r, resolvedDomains),
+	}
+
+	if route.IsV4DefaultRoute(r.Network) && route.HasV6ExitPair(id, v6Merged) {
+		network.Network = "0.0.0.0/0, ::/0"
+	}
+
+	return network
+}
+
+// findBestRoutePeer returns the peer actively routing traffic for the given
+// HA route group. Falls back to the first connected peer, then the first peer.
+func (c *Client) findBestRoutePeer(routes []*route.Route) (peer.State, error) {
+	netStr := routes[0].Network.String()
+
+	fullStatus := c.recorder.GetFullStatus()
+	for _, p := range fullStatus.Peers {
+		if _, ok := p.GetRoutes()[netStr]; ok {
+			return p, nil
+		}
+	}
+
+	for _, r := range routes {
+		p, err := c.recorder.GetPeer(r.Peer)
+		if err != nil {
+			continue
+		}
+		if p.ConnStatus == peer.StatusConnected {
+			return p, nil
+		}
+	}
+	return c.recorder.GetPeer(routes[0].Peer)
+}
+
 // OnUpdatedHostDNS update the DNS servers addresses for root zones
 func (c *Client) OnUpdatedHostDNS(list *DNSList) error {
 	dnsServer, err := dns.GetServerDns()

client/android/env_list.go

@@ -1,10 +1,19 @@
 package android
 
-import "github.com/netbirdio/netbird/client/internal/peer"
+import (
+	"github.com/netbirdio/netbird/client/internal/lazyconn"
+	"github.com/netbirdio/netbird/client/internal/peer"
+)
 
 var (
-	// EnvKeyNBForceRelay Exported for Android java client
+	// EnvKeyNBForceRelay Exported for Android java client to force relay connections
 	EnvKeyNBForceRelay = peer.EnvKeyNBForceRelay
+
+	// EnvKeyNBLazyConn Exported for Android java client to configure lazy connection
+	EnvKeyNBLazyConn = lazyconn.EnvEnableLazyConn
+
+	// EnvKeyNBInactivityThreshold Exported for Android java client to configure connection inactivity threshold
+	EnvKeyNBInactivityThreshold = lazyconn.EnvInactivityThreshold
 )
 
 // EnvList wraps a Go map for export to Java

client/android/login.go

@@ -3,15 +3,7 @@ package android
 import (
 	"context"
 	"fmt"
-	"time"
 
-	"github.com/cenkalti/backoff/v4"
-	log "github.com/sirupsen/logrus"
-	"google.golang.org/grpc/codes"
-	gstatus "google.golang.org/grpc/status"
-
-	"github.com/netbirdio/netbird/client/cmd"
-	"github.com/netbirdio/netbird/client/internal"
 	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	"github.com/netbirdio/netbird/client/system"
@@ -84,34 +76,21 @@ func (a *Auth) SaveConfigIfSSOSupported(listener SSOListener) {
 }
 
 func (a *Auth) saveConfigIfSSOSupported() (bool, error) {
-	supportsSSO := true
-	err := a.withBackOff(a.ctx, func() (err error) {
-		_, err = internal.GetPKCEAuthorizationFlowInfo(a.ctx, a.config.PrivateKey, a.config.ManagementURL, nil)
-		if s, ok := gstatus.FromError(err); ok && (s.Code() == codes.NotFound || s.Code() == codes.Unimplemented) {
-			_, err = internal.GetDeviceAuthorizationFlowInfo(a.ctx, a.config.PrivateKey, a.config.ManagementURL)
-			s, ok := gstatus.FromError(err)
-			if !ok {
-				return err
-			}
-			if s.Code() == codes.NotFound || s.Code() == codes.Unimplemented {
-				supportsSSO = false
-				err = nil
-			}
+	authClient, err := auth.NewAuth(a.ctx, a.config.PrivateKey, a.config.ManagementURL, a.config)
+	if err != nil {
+		return false, fmt.Errorf("failed to create auth client: %v", err)
+	}
+	defer authClient.Close()
 
-			return err
-		}
-
-		return err
-	})
+	supportsSSO, err := authClient.IsSSOSupported(a.ctx)
+	if err != nil {
+		return false, fmt.Errorf("failed to check SSO support: %v", err)
+	}
 
 	if !supportsSSO {
 		return false, nil
 	}
 
-	if err != nil {
-		return false, fmt.Errorf("backoff cycle failed: %v", err)
-	}
-
 	err = profilemanager.WriteOutConfig(a.cfgPath, a.config)
 	return true, err
 }
@@ -129,19 +108,17 @@ func (a *Auth) LoginWithSetupKeyAndSaveConfig(resultListener ErrListener, setupK
 }
 
 func (a *Auth) loginWithSetupKeyAndSaveConfig(setupKey string, deviceName string) error {
+	authClient, err := auth.NewAuth(a.ctx, a.config.PrivateKey, a.config.ManagementURL, a.config)
+	if err != nil {
+		return fmt.Errorf("failed to create auth client: %v", err)
+	}
+	defer authClient.Close()
+
 	//nolint
 	ctxWithValues := context.WithValue(a.ctx, system.DeviceNameCtxKey, deviceName)
-
-	err := a.withBackOff(a.ctx, func() error {
-		backoffErr := internal.Login(ctxWithValues, a.config, setupKey, "")
-		if s, ok := gstatus.FromError(backoffErr); ok && (s.Code() == codes.PermissionDenied) {
-			// we got an answer from management, exit backoff earlier
-			return backoff.Permanent(backoffErr)
-		}
-		return backoffErr
-	})
+	err, _ = authClient.Login(ctxWithValues, setupKey, "")
 	if err != nil {
-		return fmt.Errorf("backoff cycle failed: %v", err)
+		return fmt.Errorf("login failed: %v", err)
 	}
 
 	return profilemanager.WriteOutConfig(a.cfgPath, a.config)
@@ -160,49 +137,41 @@ func (a *Auth) Login(resultListener ErrListener, urlOpener URLOpener, isAndroidT
 }
 
 func (a *Auth) login(urlOpener URLOpener, isAndroidTV bool) error {
-	var needsLogin bool
+	authClient, err := auth.NewAuth(a.ctx, a.config.PrivateKey, a.config.ManagementURL, a.config)
+	if err != nil {
+		return fmt.Errorf("failed to create auth client: %v", err)
+	}
+	defer authClient.Close()
 
 	// check if we need to generate JWT token
-	err := a.withBackOff(a.ctx, func() (err error) {
-		needsLogin, err = internal.IsLoginRequired(a.ctx, a.config)
-		return
-	})
+	needsLogin, err := authClient.IsLoginRequired(a.ctx)
 	if err != nil {
-		return fmt.Errorf("backoff cycle failed: %v", err)
+		return fmt.Errorf("failed to check login requirement: %v", err)
 	}
 
 	jwtToken := ""
 	if needsLogin {
-		tokenInfo, err := a.foregroundGetTokenInfo(urlOpener, isAndroidTV)
+		tokenInfo, err := a.foregroundGetTokenInfo(authClient, urlOpener, isAndroidTV)
 		if err != nil {
 			return fmt.Errorf("interactive sso login failed: %v", err)
 		}
 		jwtToken = tokenInfo.GetTokenToUse()
 	}
 
-	err = a.withBackOff(a.ctx, func() error {
-		err := internal.Login(a.ctx, a.config, "", jwtToken)
-
-		if err == nil {
-			go urlOpener.OnLoginSuccess()
-		}
-
-		if s, ok := gstatus.FromError(err); ok && (s.Code() == codes.InvalidArgument || s.Code() == codes.PermissionDenied) {
-			return nil
-		}
-		return err
-	})
+	err, _ = authClient.Login(a.ctx, "", jwtToken)
 	if err != nil {
-		return fmt.Errorf("backoff cycle failed: %v", err)
+		return fmt.Errorf("login failed: %v", err)
 	}
 
+	go urlOpener.OnLoginSuccess()
+
 	return nil
 }
 
-func (a *Auth) foregroundGetTokenInfo(urlOpener URLOpener, isAndroidTV bool) (*auth.TokenInfo, error) {
-	oAuthFlow, err := auth.NewOAuthFlow(a.ctx, a.config, false, isAndroidTV, "")
+func (a *Auth) foregroundGetTokenInfo(authClient *auth.Auth, urlOpener URLOpener, isAndroidTV bool) (*auth.TokenInfo, error) {
+	oAuthFlow, err := authClient.GetOAuthFlow(a.ctx, isAndroidTV)
 	if err != nil {
-		return nil, err
+		return nil, fmt.Errorf("failed to get OAuth flow: %v", err)
 	}
 
 	flowInfo, err := oAuthFlow.RequestAuthInfo(context.TODO())
@@ -212,22 +181,10 @@ func (a *Auth) foregroundGetTokenInfo(urlOpener URLOpener, isAndroidTV bool) (*a
 
 	go urlOpener.Open(flowInfo.VerificationURIComplete, flowInfo.UserCode)
 
-	waitTimeout := time.Duration(flowInfo.ExpiresIn) * time.Second
-	waitCTX, cancel := context.WithTimeout(a.ctx, waitTimeout)
-	defer cancel()
-	tokenInfo, err := oAuthFlow.WaitToken(waitCTX, flowInfo)
+	tokenInfo, err := oAuthFlow.WaitToken(a.ctx, flowInfo)
 	if err != nil {
 		return nil, fmt.Errorf("waiting for browser login failed: %v", err)
 	}
 
 	return &tokenInfo, nil
 }
-
-func (a *Auth) withBackOff(ctx context.Context, bf func() error) error {
-	return backoff.RetryNotify(
-		bf,
-		backoff.WithContext(cmd.CLIBackOffSettings, ctx),
-		func(err error, duration time.Duration) {
-			log.Warnf("retrying Login to the Management service in %v due to error %v", duration, err)
-		})
-}

client/android/peer_notifier.go

@@ -5,6 +5,7 @@ package android
 // PeerInfo describe information about the peers. It designed for the UI usage
 type PeerInfo struct {
 	IP         string
+	IPv6       string
 	FQDN       string
 	ConnStatus string // Todo replace to enum
 	Routes     PeerRoutes

client/android/preferences.go

@@ -307,6 +307,24 @@ func (p *Preferences) SetBlockInbound(block bool) {
 	p.configInput.BlockInbound = &block
 }
 
+// GetDisableIPv6 reads disable IPv6 setting from config file
+func (p *Preferences) GetDisableIPv6() (bool, error) {
+	if p.configInput.DisableIPv6 != nil {
+		return *p.configInput.DisableIPv6, nil
+	}
+
+	cfg, err := profilemanager.ReadConfig(p.configInput.ConfigPath)
+	if err != nil {
+		return false, err
+	}
+	return cfg.DisableIPv6, err
+}
+
+// SetDisableIPv6 stores the given value and waits for commit
+func (p *Preferences) SetDisableIPv6(disable bool) {
+	p.configInput.DisableIPv6 = &disable
+}
+
 // Commit writes out the changes to the config file
 func (p *Preferences) Commit() error {
 	_, err := profilemanager.UpdateOrCreateConfig(p.configInput)

client/android/route_command.go

@@ -18,9 +18,12 @@ func executeRouteToggle(id string, manager routemanager.Manager,
 	netID := route.NetID(id)
 	routes := []route.NetID{netID}
 
-	log.Debugf("%s with id: %s", operationName, id)
+	routesMap := manager.GetClientRoutesWithNetID()
+	routes = route.ExpandV6ExitPairs(routes, routesMap)
 
-	if err := routeOperation(routes, maps.Keys(manager.GetClientRoutesWithNetID())); err != nil {
+	log.Debugf("%s with ids: %v", operationName, routes)
+
+	if err := routeOperation(routes, maps.Keys(routesMap)); err != nil {
 		log.Debugf("error when %s: %s", operationName, err)
 		return fmt.Errorf("error %s: %w", operationName, err)
 	}

client/anonymize/anonymize.go

@@ -9,6 +9,7 @@ import (
 	"net/url"
 	"regexp"
 	"slices"
+	"strconv"
 	"strings"
 )
 
@@ -26,8 +27,9 @@ type Anonymizer struct {
 }
 
 func DefaultAddresses() (netip.Addr, netip.Addr) {
-	// 198.51.100.0, 100::
-	return netip.AddrFrom4([4]byte{198, 51, 100, 0}), netip.AddrFrom16([16]byte{0x01})
+	// 198.51.100.0 (RFC 5737 TEST-NET-2), 2001:db8:ffff:: (RFC 3849 documentation, last /48)
+	// The old start 100:: (discard, RFC 6666) is now used for fake IPs on Android.
+	return netip.AddrFrom4([4]byte{198, 51, 100, 0}), netip.MustParseAddr("2001:db8:ffff::")
 }
 
 func NewAnonymizer(startIPv4, startIPv6 netip.Addr) *Anonymizer {
@@ -96,6 +98,11 @@ func (a *Anonymizer) isInAnonymizedRange(ip netip.Addr) bool {
 }
 
 func (a *Anonymizer) AnonymizeIPString(ip string) string {
+	// Handle CIDR notation (e.g. "2001:db8::/32")
+	if prefix, err := netip.ParsePrefix(ip); err == nil {
+		return a.AnonymizeIP(prefix.Addr()).String() + "/" + strconv.Itoa(prefix.Bits())
+	}
+
 	addr, err := netip.ParseAddr(ip)
 	if err != nil {
 		return ip

client/anonymize/anonymize_test.go

@@ -13,7 +13,7 @@ import (
 
 func TestAnonymizeIP(t *testing.T) {
 	startIPv4 := netip.MustParseAddr("198.51.100.0")
-	startIPv6 := netip.MustParseAddr("100::")
+	startIPv6 := netip.MustParseAddr("2001:db8:ffff::")
 	anonymizer := anonymize.NewAnonymizer(startIPv4, startIPv6)
 
 	tests := []struct {
@@ -26,9 +26,9 @@ func TestAnonymizeIP(t *testing.T) {
 		{"Second Public IPv4", "4.3.2.1", "198.51.100.1"},
 		{"Repeated IPv4", "1.2.3.4", "198.51.100.0"},
 		{"Private IPv4", "192.168.1.1", "192.168.1.1"},
-		{"First Public IPv6", "2607:f8b0:4005:805::200e", "100::"},
-		{"Second Public IPv6", "a::b", "100::1"},
-		{"Repeated IPv6", "2607:f8b0:4005:805::200e", "100::"},
+		{"First Public IPv6", "2607:f8b0:4005:805::200e", "2001:db8:ffff::"},
+		{"Second Public IPv6", "a::b", "2001:db8:ffff::1"},
+		{"Repeated IPv6", "2607:f8b0:4005:805::200e", "2001:db8:ffff::"},
 		{"Private IPv6", "fe80::1", "fe80::1"},
 		{"In Range IPv4", "198.51.100.2", "198.51.100.2"},
 	}
@@ -274,17 +274,17 @@ func TestAnonymizeString_IPAddresses(t *testing.T) {
 		{
 			name:   "IPv6 Address",
 			input:  "Access attempted from 2001:db8::ff00:42",
-			expect: "Access attempted from 100::",
+			expect: "Access attempted from 2001:db8:ffff::",
 		},
 		{
 			name:   "IPv6 Address with Port",
 			input:  "Access attempted from [2001:db8::ff00:42]:8080",
-			expect: "Access attempted from [100::]:8080",
+			expect: "Access attempted from [2001:db8:ffff::]:8080",
 		},
 		{
 			name:   "Both IPv4 and IPv6",
 			input:  "IPv4: 142.108.0.1 and IPv6: 2001:db8::ff00:43",
-			expect: "IPv4: 198.51.100.1 and IPv6: 100::1",
+			expect: "IPv4: 198.51.100.1 and IPv6: 2001:db8:ffff::1",
 		},
 	}
 

client/cmd/debug.go

@@ -16,7 +16,6 @@ import (
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	"github.com/netbirdio/netbird/client/proto"
 	"github.com/netbirdio/netbird/client/server"
-	nbstatus "github.com/netbirdio/netbird/client/status"
 	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
 	"github.com/netbirdio/netbird/upload-server/types"
 )
@@ -98,7 +97,6 @@ func debugBundle(cmd *cobra.Command, _ []string) error {
 	client := proto.NewDaemonServiceClient(conn)
 	request := &proto.DebugBundleRequest{
 		Anonymize:    anonymizeFlag,
-		Status:       getStatusOutput(cmd, anonymizeFlag),
 		SystemInfo:   systemInfoFlag,
 		LogFileCount: logFileCount,
 	}
@@ -136,6 +134,7 @@ func setLogLevel(cmd *cobra.Command, args []string) error {
 	client := proto.NewDaemonServiceClient(conn)
 	level := server.ParseLogLevel(args[0])
 	if level == proto.LogLevel_UNKNOWN {
+		//nolint
 		return fmt.Errorf("unknown log level: %s. Available levels are: panic, fatal, error, warn, info, debug, trace\n", args[0])
 	}
 
@@ -182,10 +181,11 @@ func runForDuration(cmd *cobra.Command, args []string) error {
 
 	if stateWasDown {
 		if _, err := client.Up(cmd.Context(), &proto.UpRequest{}); err != nil {
-			return fmt.Errorf("failed to up: %v", status.Convert(err).Message())
+			cmd.PrintErrf("Failed to bring service up: %v\n", status.Convert(err).Message())
+		} else {
+			cmd.Println("netbird up")
+			time.Sleep(time.Second * 10)
 		}
-		cmd.Println("netbird up")
-		time.Sleep(time.Second * 10)
 	}
 
 	initialLevelTrace := initialLogLevel.GetLevel() >= proto.LogLevel_TRACE
@@ -200,9 +200,10 @@ func runForDuration(cmd *cobra.Command, args []string) error {
 	}
 
 	if _, err := client.Down(cmd.Context(), &proto.DownRequest{}); err != nil {
-		return fmt.Errorf("failed to down: %v", status.Convert(err).Message())
+		cmd.PrintErrf("Failed to bring service down: %v\n", status.Convert(err).Message())
+	} else {
+		cmd.Println("netbird down")
 	}
-	cmd.Println("netbird down")
 
 	time.Sleep(1 * time.Second)
 
@@ -210,31 +211,48 @@ func runForDuration(cmd *cobra.Command, args []string) error {
 	if _, err := client.SetSyncResponsePersistence(cmd.Context(), &proto.SetSyncResponsePersistenceRequest{
 		Enabled: true,
 	}); err != nil {
-		return fmt.Errorf("failed to enable sync response persistence: %v", status.Convert(err).Message())
+		cmd.PrintErrf("Failed to enable sync response persistence: %v\n", status.Convert(err).Message())
 	}
 
 	if _, err := client.Up(cmd.Context(), &proto.UpRequest{}); err != nil {
-		return fmt.Errorf("failed to up: %v", status.Convert(err).Message())
+		cmd.PrintErrf("Failed to bring service up: %v\n", status.Convert(err).Message())
+	} else {
+		cmd.Println("netbird up")
 	}
-	cmd.Println("netbird up")
 
 	time.Sleep(3 * time.Second)
 
-	headerPostUp := fmt.Sprintf("----- NetBird post-up - Timestamp: %s", time.Now().Format(time.RFC3339))
-	statusOutput := fmt.Sprintf("%s\n%s", headerPostUp, getStatusOutput(cmd, anonymizeFlag))
+	cpuProfilingStarted := false
+	if _, err := client.StartCPUProfile(cmd.Context(), &proto.StartCPUProfileRequest{}); err != nil {
+		cmd.PrintErrf("Failed to start CPU profiling: %v\n", err)
+	} else {
+		cpuProfilingStarted = true
+		defer func() {
+			if cpuProfilingStarted {
+				if _, err := client.StopCPUProfile(cmd.Context(), &proto.StopCPUProfileRequest{}); err != nil {
+					cmd.PrintErrf("Failed to stop CPU profiling: %v\n", err)
+				}
+			}
+		}()
+	}
 
 	if waitErr := waitForDurationOrCancel(cmd.Context(), duration, cmd); waitErr != nil {
 		return waitErr
 	}
 	cmd.Println("\nDuration completed")
 
+	if cpuProfilingStarted {
+		if _, err := client.StopCPUProfile(cmd.Context(), &proto.StopCPUProfileRequest{}); err != nil {
+			cmd.PrintErrf("Failed to stop CPU profiling: %v\n", err)
+		} else {
+			cpuProfilingStarted = false
+		}
+	}
+
 	cmd.Println("Creating debug bundle...")
 
-	headerPreDown := fmt.Sprintf("----- NetBird pre-down - Timestamp: %s - Duration: %s", time.Now().Format(time.RFC3339), duration)
-	statusOutput = fmt.Sprintf("%s\n%s\n%s", statusOutput, headerPreDown, getStatusOutput(cmd, anonymizeFlag))
 	request := &proto.DebugBundleRequest{
 		Anonymize:    anonymizeFlag,
-		Status:       statusOutput,
 		SystemInfo:   systemInfoFlag,
 		LogFileCount: logFileCount,
 	}
@@ -248,16 +266,18 @@ func runForDuration(cmd *cobra.Command, args []string) error {
 
 	if stateWasDown {
 		if _, err := client.Down(cmd.Context(), &proto.DownRequest{}); err != nil {
-			return fmt.Errorf("failed to down: %v", status.Convert(err).Message())
+			cmd.PrintErrf("Failed to restore service down state: %v\n", status.Convert(err).Message())
+		} else {
+			cmd.Println("netbird down")
 		}
-		cmd.Println("netbird down")
 	}
 
 	if !initialLevelTrace {
 		if _, err := client.SetLogLevel(cmd.Context(), &proto.SetLogLevelRequest{Level: initialLogLevel.GetLevel()}); err != nil {
-			return fmt.Errorf("failed to restore log level: %v", status.Convert(err).Message())
+			cmd.PrintErrf("Failed to restore log level: %v\n", status.Convert(err).Message())
+		} else {
+			cmd.Println("Log level restored to", initialLogLevel.GetLevel())
 		}
-		cmd.Println("Log level restored to", initialLogLevel.GetLevel())
 	}
 
 	cmd.Printf("Local file:\n%s\n", resp.GetPath())
@@ -301,25 +321,6 @@ func setSyncResponsePersistence(cmd *cobra.Command, args []string) error {
 	return nil
 }
 
-func getStatusOutput(cmd *cobra.Command, anon bool) string {
-	var statusOutputString string
-	statusResp, err := getStatus(cmd.Context(), true)
-	if err != nil {
-		cmd.PrintErrf("Failed to get status: %v\n", err)
-	} else {
-		pm := profilemanager.NewProfileManager()
-		var profName string
-		if activeProf, err := pm.GetActiveProfile(); err == nil {
-			profName = activeProf.Name
-		}
-
-		statusOutputString = nbstatus.ParseToFullDetailSummary(
-			nbstatus.ConvertToStatusOutputOverview(statusResp, anon, "", nil, nil, nil, "", profName),
-		)
-	}
-	return statusOutputString
-}
-
 func waitForDurationOrCancel(ctx context.Context, duration time.Duration, cmd *cobra.Command) error {
 	ticker := time.NewTicker(1 * time.Second)
 	defer ticker.Stop()
@@ -378,7 +379,8 @@ func generateDebugBundle(config *profilemanager.Config, recorder *peer.Status, c
 			InternalConfig: config,
 			StatusRecorder: recorder,
 			SyncResponse:   syncResponse,
-			LogFile:        logFilePath,
+			LogPath:        logFilePath,
+			CPUProfile:     nil,
 		},
 		debug.BundleConfig{
 			IncludeSystemInfo: true,

client/cmd/expose.go

@@ -0,0 +1,280 @@
+package cmd
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"io"
+	"os"
+	"os/signal"
+	"regexp"
+	"strconv"
+	"strings"
+	"syscall"
+
+	log "github.com/sirupsen/logrus"
+	"github.com/spf13/cobra"
+
+	"github.com/netbirdio/netbird/client/proto"
+	"github.com/netbirdio/netbird/util"
+)
+
+var pinRegexp = regexp.MustCompile(`^\d{6}$`)
+
+var (
+	exposePin          string
+	exposePassword     string
+	exposeUserGroups   []string
+	exposeDomain       string
+	exposeNamePrefix   string
+	exposeProtocol     string
+	exposeExternalPort uint16
+)
+
+var exposeCmd = &cobra.Command{
+	Use:   "expose <port>",
+	Short: "Expose a local port via the NetBird reverse proxy",
+	Args:  cobra.ExactArgs(1),
+	Example: `  netbird expose --with-password safe-pass 8080
+  netbird expose --protocol tcp 5432
+  netbird expose --protocol tcp --with-external-port 5433 5432
+  netbird expose --protocol tls --with-custom-domain tls.example.com 4443`,
+	RunE: exposeFn,
+}
+
+func init() {
+	exposeCmd.Flags().StringVar(&exposePin, "with-pin", "", "Protect the exposed service with a 6-digit PIN (e.g. --with-pin 123456)")
+	exposeCmd.Flags().StringVar(&exposePassword, "with-password", "", "Protect the exposed service with a password (e.g. --with-password my-secret)")
+	exposeCmd.Flags().StringSliceVar(&exposeUserGroups, "with-user-groups", nil, "Restrict access to specific user groups with SSO (e.g. --with-user-groups devops,Backend)")
+	exposeCmd.Flags().StringVar(&exposeDomain, "with-custom-domain", "", "Custom domain for the exposed service, must be configured to your account (e.g. --with-custom-domain myapp.example.com)")
+	exposeCmd.Flags().StringVar(&exposeNamePrefix, "with-name-prefix", "", "Prefix for the generated service name (e.g. --with-name-prefix my-app)")
+	exposeCmd.Flags().StringVar(&exposeProtocol, "protocol", "http", "Protocol to use: http, https, tcp, udp, or tls (e.g. --protocol tcp)")
+	exposeCmd.Flags().Uint16Var(&exposeExternalPort, "with-external-port", 0, "Public-facing external port on the proxy cluster (defaults to the target port for L4)")
+}
+
+// isClusterProtocol returns true for L4/TLS protocols that reject HTTP-style auth flags.
+func isClusterProtocol(protocol string) bool {
+	switch strings.ToLower(protocol) {
+	case "tcp", "udp", "tls":
+		return true
+	default:
+		return false
+	}
+}
+
+// isPortBasedProtocol returns true for pure port-based protocols (TCP/UDP)
+// where domain display doesn't apply. TLS uses SNI so it has a domain.
+func isPortBasedProtocol(protocol string) bool {
+	switch strings.ToLower(protocol) {
+	case "tcp", "udp":
+		return true
+	default:
+		return false
+	}
+}
+
+// extractPort returns the port portion of a URL like "tcp://host:12345", or
+// falls back to the given default formatted as a string.
+func extractPort(serviceURL string, fallback uint16) string {
+	u := serviceURL
+	if idx := strings.Index(u, "://"); idx != -1 {
+		u = u[idx+3:]
+	}
+	if i := strings.LastIndex(u, ":"); i != -1 {
+		if p := u[i+1:]; p != "" {
+			return p
+		}
+	}
+	return strconv.FormatUint(uint64(fallback), 10)
+}
+
+// resolveExternalPort returns the effective external port, defaulting to the target port.
+func resolveExternalPort(targetPort uint64) uint16 {
+	if exposeExternalPort != 0 {
+		return exposeExternalPort
+	}
+	return uint16(targetPort)
+}
+
+func validateExposeFlags(cmd *cobra.Command, portStr string) (uint64, error) {
+	port, err := strconv.ParseUint(portStr, 10, 32)
+	if err != nil {
+		return 0, fmt.Errorf("invalid port number: %s", portStr)
+	}
+	if port == 0 || port > 65535 {
+		return 0, fmt.Errorf("invalid port number: must be between 1 and 65535")
+	}
+
+	if !isProtocolValid(exposeProtocol) {
+		return 0, fmt.Errorf("unsupported protocol %q: must be http, https, tcp, udp, or tls", exposeProtocol)
+	}
+
+	if isClusterProtocol(exposeProtocol) {
+		if exposePin != "" || exposePassword != "" || len(exposeUserGroups) > 0 {
+			return 0, fmt.Errorf("auth flags (--with-pin, --with-password, --with-user-groups) are not supported for %s protocol", exposeProtocol)
+		}
+	} else if cmd.Flags().Changed("with-external-port") {
+		return 0, fmt.Errorf("--with-external-port is not supported for %s protocol", exposeProtocol)
+	}
+
+	if exposePin != "" && !pinRegexp.MatchString(exposePin) {
+		return 0, fmt.Errorf("invalid pin: must be exactly 6 digits")
+	}
+
+	if cmd.Flags().Changed("with-password") && exposePassword == "" {
+		return 0, fmt.Errorf("password cannot be empty")
+	}
+
+	if cmd.Flags().Changed("with-user-groups") && len(exposeUserGroups) == 0 {
+		return 0, fmt.Errorf("user groups cannot be empty")
+	}
+
+	return port, nil
+}
+
+func isProtocolValid(exposeProtocol string) bool {
+	switch strings.ToLower(exposeProtocol) {
+	case "http", "https", "tcp", "udp", "tls":
+		return true
+	default:
+		return false
+	}
+}
+
+func exposeFn(cmd *cobra.Command, args []string) error {
+	SetFlagsFromEnvVars(rootCmd)
+
+	if err := util.InitLog(logLevel, util.LogConsole); err != nil {
+		log.Errorf("failed initializing log %v", err)
+		return err
+	}
+
+	cmd.Root().SilenceUsage = false
+
+	port, err := validateExposeFlags(cmd, args[0])
+	if err != nil {
+		return err
+	}
+
+	cmd.Root().SilenceUsage = true
+
+	ctx, cancel := context.WithCancel(cmd.Context())
+	defer cancel()
+
+	sigCh := make(chan os.Signal, 1)
+	signal.Notify(sigCh, syscall.SIGINT, syscall.SIGTERM)
+	go func() {
+		<-sigCh
+		cancel()
+	}()
+
+	conn, err := DialClientGRPCServer(ctx, daemonAddr)
+	if err != nil {
+		return fmt.Errorf("connect to daemon: %w", err)
+	}
+	defer func() {
+		if err := conn.Close(); err != nil {
+			log.Debugf("failed to close daemon connection: %v", err)
+		}
+	}()
+
+	client := proto.NewDaemonServiceClient(conn)
+
+	protocol, err := toExposeProtocol(exposeProtocol)
+	if err != nil {
+		return err
+	}
+
+	req := &proto.ExposeServiceRequest{
+		Port:       uint32(port),
+		Protocol:   protocol,
+		Pin:        exposePin,
+		Password:   exposePassword,
+		UserGroups: exposeUserGroups,
+		Domain:     exposeDomain,
+		NamePrefix: exposeNamePrefix,
+	}
+	if isClusterProtocol(exposeProtocol) {
+		req.ListenPort = uint32(resolveExternalPort(port))
+	}
+
+	stream, err := client.ExposeService(ctx, req)
+	if err != nil {
+		return fmt.Errorf("expose service: %w", err)
+	}
+
+	if err := handleExposeReady(cmd, stream, port); err != nil {
+		return err
+	}
+
+	return waitForExposeEvents(cmd, ctx, stream)
+}
+
+func toExposeProtocol(exposeProtocol string) (proto.ExposeProtocol, error) {
+	switch strings.ToLower(exposeProtocol) {
+	case "http":
+		return proto.ExposeProtocol_EXPOSE_HTTP, nil
+	case "https":
+		return proto.ExposeProtocol_EXPOSE_HTTPS, nil
+	case "tcp":
+		return proto.ExposeProtocol_EXPOSE_TCP, nil
+	case "udp":
+		return proto.ExposeProtocol_EXPOSE_UDP, nil
+	case "tls":
+		return proto.ExposeProtocol_EXPOSE_TLS, nil
+	default:
+		return 0, fmt.Errorf("unsupported protocol %q: must be http, https, tcp, udp, or tls", exposeProtocol)
+	}
+}
+
+func handleExposeReady(cmd *cobra.Command, stream proto.DaemonService_ExposeServiceClient, port uint64) error {
+	event, err := stream.Recv()
+	if err != nil {
+		return fmt.Errorf("receive expose event: %w", err)
+	}
+
+	ready, ok := event.Event.(*proto.ExposeServiceEvent_Ready)
+	if !ok {
+		return fmt.Errorf("unexpected expose event: %T", event.Event)
+	}
+	printExposeReady(cmd, ready.Ready, port)
+	return nil
+}
+
+func printExposeReady(cmd *cobra.Command, r *proto.ExposeServiceReady, port uint64) {
+	cmd.Println("Service exposed successfully!")
+	cmd.Printf("  Name:     %s\n", r.ServiceName)
+	if r.ServiceUrl != "" {
+		cmd.Printf("  URL:      %s\n", r.ServiceUrl)
+	}
+	if r.Domain != "" && !isPortBasedProtocol(exposeProtocol) {
+		cmd.Printf("  Domain:   %s\n", r.Domain)
+	}
+	cmd.Printf("  Protocol: %s\n", exposeProtocol)
+	cmd.Printf("  Internal: %d\n", port)
+	if isClusterProtocol(exposeProtocol) {
+		cmd.Printf("  External: %s\n", extractPort(r.ServiceUrl, resolveExternalPort(port)))
+	}
+	if r.PortAutoAssigned && exposeExternalPort != 0 {
+		cmd.Printf("\n  Note: requested port %d was reassigned\n", exposeExternalPort)
+	}
+	cmd.Println()
+	cmd.Println("Press Ctrl+C to stop exposing.")
+}
+
+func waitForExposeEvents(cmd *cobra.Command, ctx context.Context, stream proto.DaemonService_ExposeServiceClient) error {
+	for {
+		_, err := stream.Recv()
+		if err != nil {
+			if ctx.Err() != nil {
+				cmd.Println("\nService stopped.")
+				//nolint:nilerr
+				return nil
+			}
+			if errors.Is(err, io.EOF) {
+				return fmt.Errorf("connection to daemon closed unexpectedly")
+			}
+			return fmt.Errorf("stream error: %w", err)
+		}
+	}
+}

client/cmd/login.go

@@ -7,7 +7,6 @@ import (
 	"os/user"
 	"runtime"
 	"strings"
-	"time"
 
 	log "github.com/sirupsen/logrus"
 	"github.com/spf13/cobra"
@@ -81,6 +80,7 @@ var loginCmd = &cobra.Command{
 func doDaemonLogin(ctx context.Context, cmd *cobra.Command, providedSetupKey string, activeProf *profilemanager.Profile, username string, pm *profilemanager.ProfileManager) error {
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
+		//nolint
 		return fmt.Errorf("failed to connect to daemon error: %v\n"+
 			"If the daemon is not running please run: "+
 			"\nnetbird service install \nnetbird service start\n", err)
@@ -206,6 +206,7 @@ func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManage
 func switchProfile(ctx context.Context, profileName string, username string) error {
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
+		//nolint
 		return fmt.Errorf("failed to connect to daemon error: %v\n"+
 			"If the daemon is not running please run: "+
 			"\nnetbird service install \nnetbird service start\n", err)
@@ -275,18 +276,15 @@ func handleSSOLogin(ctx context.Context, cmd *cobra.Command, loginResp *proto.Lo
 }
 
 func foregroundLogin(ctx context.Context, cmd *cobra.Command, config *profilemanager.Config, setupKey, profileName string) error {
-	needsLogin := false
-
-	err := WithBackOff(func() error {
-		err := internal.Login(ctx, config, "", "")
-		if s, ok := gstatus.FromError(err); ok && (s.Code() == codes.InvalidArgument || s.Code() == codes.PermissionDenied) {
-			needsLogin = true
-			return nil
-		}
-		return err
-	})
+	authClient, err := auth.NewAuth(ctx, config.PrivateKey, config.ManagementURL, config)
 	if err != nil {
-		return fmt.Errorf("backoff cycle failed: %v", err)
+		return fmt.Errorf("failed to create auth client: %v", err)
+	}
+	defer authClient.Close()
+
+	needsLogin, err := authClient.IsLoginRequired(ctx)
+	if err != nil {
+		return fmt.Errorf("check login required: %v", err)
 	}
 
 	jwtToken := ""
@@ -298,23 +296,9 @@ func foregroundLogin(ctx context.Context, cmd *cobra.Command, config *profileman
 		jwtToken = tokenInfo.GetTokenToUse()
 	}
 
-	var lastError error
-
-	err = WithBackOff(func() error {
-		err := internal.Login(ctx, config, setupKey, jwtToken)
-		if s, ok := gstatus.FromError(err); ok && (s.Code() == codes.InvalidArgument || s.Code() == codes.PermissionDenied) {
-			lastError = err
-			return nil
-		}
-		return err
-	})
-
-	if lastError != nil {
-		return fmt.Errorf("login failed: %v", lastError)
-	}
-
+	err, _ = authClient.Login(ctx, setupKey, jwtToken)
 	if err != nil {
-		return fmt.Errorf("backoff cycle failed: %v", err)
+		return fmt.Errorf("login failed: %v", err)
 	}
 
 	return nil
@@ -342,11 +326,7 @@ func foregroundGetTokenInfo(ctx context.Context, cmd *cobra.Command, config *pro
 
 	openURL(cmd, flowInfo.VerificationURIComplete, flowInfo.UserCode, noBrowser)
 
-	waitTimeout := time.Duration(flowInfo.ExpiresIn) * time.Second
-	waitCTX, c := context.WithTimeout(context.TODO(), waitTimeout)
-	defer c()
-
-	tokenInfo, err := oAuthFlow.WaitToken(waitCTX, flowInfo)
+	tokenInfo, err := oAuthFlow.WaitToken(context.TODO(), flowInfo)
 	if err != nil {
 		return nil, fmt.Errorf("waiting for browser login failed: %v", err)
 	}

client/cmd/pprof.go

@@ -1,5 +1,4 @@
 //go:build pprof
-// +build pprof
 
 package cmd
 

client/cmd/root.go

@@ -22,6 +22,7 @@ import (
 	"google.golang.org/grpc"
 	"google.golang.org/grpc/credentials/insecure"
 
+	daddr "github.com/netbirdio/netbird/client/internal/daemonaddr"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 )
 
@@ -80,6 +81,15 @@ var (
 		Short:        "",
 		Long:         "",
 		SilenceUsage: true,
+		PersistentPreRunE: func(cmd *cobra.Command, args []string) error {
+			SetFlagsFromEnvVars(cmd.Root())
+
+			// Don't resolve for service commands — they create the socket, not connect to it.
+			if !isServiceCmd(cmd) {
+				daemonAddr = daddr.ResolveUnixDaemonAddr(daemonAddr)
+			}
+			return nil
+		},
 	}
 )
 
@@ -144,6 +154,7 @@ func init() {
 	rootCmd.AddCommand(forwardingRulesCmd)
 	rootCmd.AddCommand(debugCmd)
 	rootCmd.AddCommand(profileCmd)
+	rootCmd.AddCommand(exposeCmd)
 
 	networksCMD.AddCommand(routesListCmd)
 	networksCMD.AddCommand(routesSelectCmd, routesDeselectCmd)
@@ -385,11 +396,11 @@ func migrateToNetbird(oldPath, newPath string) bool {
 }
 
 func getClient(cmd *cobra.Command) (*grpc.ClientConn, error) {
-	SetFlagsFromEnvVars(rootCmd)
 	cmd.SetOut(cmd.OutOrStdout())
 
 	conn, err := DialClientGRPCServer(cmd.Context(), daemonAddr)
 	if err != nil {
+		//nolint
 		return nil, fmt.Errorf("failed to connect to daemon error: %v\n"+
 			"If the daemon is not running please run: "+
 			"\nnetbird service install \nnetbird service start\n", err)
@@ -397,3 +408,13 @@ func getClient(cmd *cobra.Command) (*grpc.ClientConn, error) {
 
 	return conn, nil
 }
+
+// isServiceCmd returns true if cmd is the "service" command or a child of it.
+func isServiceCmd(cmd *cobra.Command) bool {
+	for c := cmd; c != nil; c = c.Parent() {
+		if c.Name() == "service" {
+			return true
+		}
+	}
+	return false
+}

client/cmd/service_controller.go

@@ -103,7 +103,7 @@ func (p *program) Stop(srv service.Service) error {
 
 // Common setup for service control commands
 func setupServiceControlCommand(cmd *cobra.Command, ctx context.Context, cancel context.CancelFunc) (service.Service, error) {
-	SetFlagsFromEnvVars(rootCmd)
+	// rootCmd env vars are already applied by PersistentPreRunE.
 	SetFlagsFromEnvVars(serviceCmd)
 
 	cmd.SetOut(cmd.OutOrStdout())

client/cmd/signer/artifactkey.go

@@ -7,7 +7,7 @@ import (
 
 	"github.com/spf13/cobra"
 
-	"github.com/netbirdio/netbird/client/internal/updatemanager/reposign"
+	"github.com/netbirdio/netbird/client/internal/updater/reposign"
 )
 
 var (

client/cmd/signer/artifactsign.go

@@ -6,7 +6,7 @@ import (
 
 	"github.com/spf13/cobra"
 
-	"github.com/netbirdio/netbird/client/internal/updatemanager/reposign"
+	"github.com/netbirdio/netbird/client/internal/updater/reposign"
 )
 
 const (

client/cmd/signer/revocation.go

@@ -7,7 +7,7 @@ import (
 
 	"github.com/spf13/cobra"
 
-	"github.com/netbirdio/netbird/client/internal/updatemanager/reposign"
+	"github.com/netbirdio/netbird/client/internal/updater/reposign"
 )
 
 const (

client/cmd/signer/rootkey.go

@@ -7,7 +7,7 @@ import (
 
 	"github.com/spf13/cobra"
 
-	"github.com/netbirdio/netbird/client/internal/updatemanager/reposign"
+	"github.com/netbirdio/netbird/client/internal/updater/reposign"
 )
 
 var (

client/cmd/ssh.go

@@ -787,10 +787,10 @@ func isUnixSocket(path string) bool {
 	return strings.HasPrefix(path, "/") || strings.HasPrefix(path, "./")
 }
 
-// normalizeLocalHost converts "*" to "0.0.0.0" for binding to all interfaces.
+// normalizeLocalHost converts "*" to "" for binding to all interfaces (dual-stack).
 func normalizeLocalHost(host string) string {
 	if host == "*" {
-		return "0.0.0.0"
+		return ""
 	}
 	return host
 }

client/cmd/ssh_test.go

@@ -527,10 +527,10 @@ func TestParsePortForward(t *testing.T) {
 		{
 			name:           "wildcard bind all interfaces",
 			spec:           "*:8080:localhost:80",
-			expectedLocal:  "0.0.0.0:8080",
+			expectedLocal:  ":8080",
 			expectedRemote: "localhost:80",
 			expectError:    false,
-			description:    "Wildcard * should bind to all interfaces (0.0.0.0)",
+			description:    "Wildcard * should bind to all interfaces (dual-stack)",
 		},
 		{
 			name:           "wildcard for port only",

client/cmd/status.go

@@ -20,6 +20,7 @@ import (
 var (
 	detailFlag           bool
 	ipv4Flag             bool
+	ipv6Flag             bool
 	jsonFlag             bool
 	yamlFlag             bool
 	ipsFilter            []string
@@ -28,6 +29,7 @@ var (
 	ipsFilterMap         map[string]struct{}
 	prefixNamesFilterMap map[string]struct{}
 	connectionTypeFilter string
+	checkFlag            string
 )
 
 var statusCmd = &cobra.Command{
@@ -44,11 +46,13 @@ func init() {
 	statusCmd.PersistentFlags().BoolVar(&jsonFlag, "json", false, "display detailed status information in json format")
 	statusCmd.PersistentFlags().BoolVar(&yamlFlag, "yaml", false, "display detailed status information in yaml format")
 	statusCmd.PersistentFlags().BoolVar(&ipv4Flag, "ipv4", false, "display only NetBird IPv4 of this peer, e.g., --ipv4 will output 100.64.0.33")
-	statusCmd.MarkFlagsMutuallyExclusive("detail", "json", "yaml", "ipv4")
-	statusCmd.PersistentFlags().StringSliceVar(&ipsFilter, "filter-by-ips", []string{}, "filters the detailed output by a list of one or more IPs, e.g., --filter-by-ips 100.64.0.100,100.64.0.200")
+	statusCmd.PersistentFlags().BoolVar(&ipv6Flag, "ipv6", false, "display only NetBird IPv6 of this peer")
+	statusCmd.MarkFlagsMutuallyExclusive("detail", "json", "yaml", "ipv4", "ipv6")
+	statusCmd.PersistentFlags().StringSliceVar(&ipsFilter, "filter-by-ips", []string{}, "filters the detailed output by a list of one or more IPs (v4 or v6), e.g., --filter-by-ips 100.64.0.100,fd00::1")
 	statusCmd.PersistentFlags().StringSliceVar(&prefixNamesFilter, "filter-by-names", []string{}, "filters the detailed output by a list of one or more peer FQDN or hostnames, e.g., --filter-by-names peer-a,peer-b.netbird.cloud")
 	statusCmd.PersistentFlags().StringVar(&statusFilter, "filter-by-status", "", "filters the detailed output by connection status(idle|connecting|connected), e.g., --filter-by-status connected")
 	statusCmd.PersistentFlags().StringVar(&connectionTypeFilter, "filter-by-connection-type", "", "filters the detailed output by connection type (P2P|Relayed), e.g., --filter-by-connection-type P2P")
+	statusCmd.PersistentFlags().StringVar(&checkFlag, "check", "", "run a health check and exit with code 0 on success, 1 on failure (live|ready|startup)")
 }
 
 func statusFunc(cmd *cobra.Command, args []string) error {
@@ -56,6 +60,10 @@ func statusFunc(cmd *cobra.Command, args []string) error {
 
 	cmd.SetOut(cmd.OutOrStdout())
 
+	if checkFlag != "" {
+		return runHealthCheck(cmd)
+	}
+
 	err := parseFilters()
 	if err != nil {
 		return err
@@ -68,15 +76,17 @@ func statusFunc(cmd *cobra.Command, args []string) error {
 
 	ctx := internal.CtxInitState(cmd.Context())
 
-	resp, err := getStatus(ctx, false)
+	resp, err := getStatus(ctx, true, false)
 	if err != nil {
 		return err
 	}
 
 	status := resp.GetStatus()
 
-	if status == string(internal.StatusNeedsLogin) || status == string(internal.StatusLoginFailed) ||
-		status == string(internal.StatusSessionExpired) {
+	needsAuth := status == string(internal.StatusNeedsLogin) || status == string(internal.StatusLoginFailed) ||
+		status == string(internal.StatusSessionExpired)
+
+	if needsAuth && !jsonFlag && !yamlFlag {
 		cmd.Printf("Daemon status: %s\n\n"+
 			"Run UP command to log in with SSO (interactive login):\n\n"+
 			" netbird up \n\n"+
@@ -93,23 +103,41 @@ func statusFunc(cmd *cobra.Command, args []string) error {
 		return nil
 	}
 
+	if ipv6Flag {
+		ipv6 := resp.GetFullStatus().GetLocalPeerState().GetIpv6()
+		if ipv6 != "" {
+			cmd.Print(parseInterfaceIP(ipv6))
+		}
+		return nil
+	}
+
 	pm := profilemanager.NewProfileManager()
 	var profName string
 	if activeProf, err := pm.GetActiveProfile(); err == nil {
 		profName = activeProf.Name
 	}
 
-	var outputInformationHolder = nbstatus.ConvertToStatusOutputOverview(resp, anonymizeFlag, statusFilter, prefixNamesFilter, prefixNamesFilterMap, ipsFilterMap, connectionTypeFilter, profName)
+	var outputInformationHolder = nbstatus.ConvertToStatusOutputOverview(resp.GetFullStatus(), nbstatus.ConvertOptions{
+		Anonymize:            anonymizeFlag,
+		DaemonVersion:        resp.GetDaemonVersion(),
+		DaemonStatus:         nbstatus.ParseDaemonStatus(status),
+		StatusFilter:         statusFilter,
+		PrefixNamesFilter:    prefixNamesFilter,
+		PrefixNamesFilterMap: prefixNamesFilterMap,
+		IPsFilter:            ipsFilterMap,
+		ConnectionTypeFilter: connectionTypeFilter,
+		ProfileName:          profName,
+	})
 	var statusOutputString string
 	switch {
 	case detailFlag:
-		statusOutputString = nbstatus.ParseToFullDetailSummary(outputInformationHolder)
+		statusOutputString = outputInformationHolder.FullDetailSummary()
 	case jsonFlag:
-		statusOutputString, err = nbstatus.ParseToJSON(outputInformationHolder)
+		statusOutputString, err = outputInformationHolder.JSON()
 	case yamlFlag:
-		statusOutputString, err = nbstatus.ParseToYAML(outputInformationHolder)
+		statusOutputString, err = outputInformationHolder.YAML()
 	default:
-		statusOutputString = nbstatus.ParseGeneralSummary(outputInformationHolder, false, false, false, false)
+		statusOutputString = outputInformationHolder.GeneralSummary(false, false, false, false)
 	}
 
 	if err != nil {
@@ -121,16 +149,17 @@ func statusFunc(cmd *cobra.Command, args []string) error {
 	return nil
 }
 
-func getStatus(ctx context.Context, shouldRunProbes bool) (*proto.StatusResponse, error) {
+func getStatus(ctx context.Context, fullPeerStatus bool, shouldRunProbes bool) (*proto.StatusResponse, error) {
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
+		//nolint
 		return nil, fmt.Errorf("failed to connect to daemon error: %v\n"+
 			"If the daemon is not running please run: "+
 			"\nnetbird service install \nnetbird service start\n", err)
 	}
 	defer conn.Close()
 
-	resp, err := proto.NewDaemonServiceClient(conn).Status(ctx, &proto.StatusRequest{GetFullPeerStatus: true, ShouldRunProbes: shouldRunProbes})
+	resp, err := proto.NewDaemonServiceClient(conn).Status(ctx, &proto.StatusRequest{GetFullPeerStatus: fullPeerStatus, ShouldRunProbes: shouldRunProbes})
 	if err != nil {
 		return nil, fmt.Errorf("status failed: %v", status.Convert(err).Message())
 	}
@@ -184,6 +213,83 @@ func enableDetailFlagWhenFilterFlag() {
 	}
 }
 
+func runHealthCheck(cmd *cobra.Command) error {
+	check := strings.ToLower(checkFlag)
+	switch check {
+	case "live", "ready", "startup":
+	default:
+		return fmt.Errorf("unknown check %q, must be one of: live, ready, startup", checkFlag)
+	}
+
+	if err := util.InitLog(logLevel, util.LogConsole); err != nil {
+		return fmt.Errorf("init log: %w", err)
+	}
+
+	ctx := internal.CtxInitState(cmd.Context())
+
+	isStartup := check == "startup"
+	resp, err := getStatus(ctx, isStartup, false)
+	if err != nil {
+		return err
+	}
+
+	switch check {
+	case "live":
+		return nil
+	case "ready":
+		return checkReadiness(resp)
+	case "startup":
+		return checkStartup(resp)
+	default:
+		return nil
+	}
+}
+
+func checkReadiness(resp *proto.StatusResponse) error {
+	daemonStatus := internal.StatusType(resp.GetStatus())
+	switch daemonStatus {
+	case internal.StatusIdle, internal.StatusConnecting, internal.StatusConnected:
+		return nil
+	case internal.StatusNeedsLogin, internal.StatusLoginFailed, internal.StatusSessionExpired:
+		return fmt.Errorf("readiness check: daemon status is %s", daemonStatus)
+	default:
+		return fmt.Errorf("readiness check: unexpected daemon status %q", daemonStatus)
+	}
+}
+
+func checkStartup(resp *proto.StatusResponse) error {
+	fullStatus := resp.GetFullStatus()
+	if fullStatus == nil {
+		return fmt.Errorf("startup check: no full status available")
+	}
+
+	if !fullStatus.GetManagementState().GetConnected() {
+		return fmt.Errorf("startup check: management not connected")
+	}
+
+	if !fullStatus.GetSignalState().GetConnected() {
+		return fmt.Errorf("startup check: signal not connected")
+	}
+
+	var relayCount, relaysConnected int
+	for _, r := range fullStatus.GetRelays() {
+		uri := r.GetURI()
+		if !strings.HasPrefix(uri, "rel://") && !strings.HasPrefix(uri, "rels://") {
+			continue
+		}
+		relayCount++
+		if r.GetAvailable() {
+			relaysConnected++
+		}
+	}
+
+	if relayCount > 0 && relaysConnected == 0 {
+		return fmt.Errorf("startup check: no relay servers available (0/%d connected)", relayCount)
+	}
+
+	return nil
+}
+
 func parseInterfaceIP(interfaceIP string) string {
 	ip, _, err := net.ParseCIDR(interfaceIP)
 	if err != nil {

client/cmd/system.go

@@ -8,6 +8,7 @@ const (
 	disableFirewallFlag     = "disable-firewall"
 	blockLANAccessFlag      = "block-lan-access"
 	blockInboundFlag        = "block-inbound"
+	disableIPv6Flag         = "disable-ipv6"
 )
 
 var (
@@ -17,6 +18,7 @@ var (
 	disableFirewall     bool
 	blockLANAccess      bool
 	blockInbound        bool
+	disableIPv6         bool
 )
 
 func init() {
@@ -39,4 +41,7 @@ func init() {
 	upCmd.PersistentFlags().BoolVar(&blockInbound, blockInboundFlag, false,
 		"Block inbound connections. If enabled, the client will not allow any inbound connections to the local machine nor routed networks.\n"+
 			"This overrides any policies received from the management service.")
+
+	upCmd.PersistentFlags().BoolVar(&disableIPv6, disableIPv6Flag, false,
+		"Disable IPv6 overlay. If enabled, the client won't request or use an IPv6 overlay address.")
 }

client/cmd/testutil_test.go

@@ -18,6 +18,7 @@ import (
 	"github.com/netbirdio/netbird/management/internals/modules/peers"
 	"github.com/netbirdio/netbird/management/internals/modules/peers/ephemeral/manager"
 	nbgrpc "github.com/netbirdio/netbird/management/internals/shared/grpc"
+	"github.com/netbirdio/netbird/management/server/job"
 
 	clientProto "github.com/netbirdio/netbird/client/proto"
 	client "github.com/netbirdio/netbird/client/server"
@@ -89,9 +90,6 @@ func startManagement(t *testing.T, config *config.Config, testFile string) (*grp
 	t.Cleanup(cleanUp)
 
 	eventStore := &activity.InMemoryEventStore{}
-	if err != nil {
-		return nil, nil
-	}
 
 	ctrl := gomock.NewController(t)
 	t.Cleanup(ctrl.Finish)
@@ -100,6 +98,8 @@ func startManagement(t *testing.T, config *config.Config, testFile string) (*grp
 	peersmanager := peers.NewManager(store, permissionsManagerMock)
 	settingsManagerMock := settings.NewMockManager(ctrl)
 
+	jobManager := job.NewJobManager(nil, store, peersmanager)
+
 	iv, _ := integrations.NewIntegratedValidator(context.Background(), peersmanager, settingsManagerMock, eventStore)
 
 	metrics, err := telemetry.NewDefaultAppMetrics(context.Background())
@@ -118,7 +118,7 @@ func startManagement(t *testing.T, config *config.Config, testFile string) (*grp
 	requestBuffer := mgmt.NewAccountRequestBuffer(ctx, store)
 	networkMapController := controller.NewController(ctx, store, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.cloud", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersmanager), config)
 
-	accountManager, err := mgmt.BuildManager(context.Background(), config, store, networkMapController, nil, "", eventStore, nil, false, iv, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManagerMock, false)
+	accountManager, err := mgmt.BuildManager(context.Background(), config, store, networkMapController, jobManager, nil, "", eventStore, nil, false, iv, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManagerMock, false)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -127,7 +127,7 @@ func startManagement(t *testing.T, config *config.Config, testFile string) (*grp
 	if err != nil {
 		t.Fatal(err)
 	}
-	mgmtServer, err := nbgrpc.NewServer(config, accountManager, settingsMockManager, secretsManager, nil, nil, &mgmt.MockIntegratedValidator{}, networkMapController, nil)
+	mgmtServer, err := nbgrpc.NewServer(config, accountManager, settingsMockManager, jobManager, secretsManager, nil, nil, &mgmt.MockIntegratedValidator{}, networkMapController, nil)
 	if err != nil {
 		t.Fatal(err)
 	}

client/cmd/up.go

@@ -197,10 +197,10 @@ func runInForegroundMode(ctx context.Context, cmd *cobra.Command, activeProf *pr
 	r := peer.NewRecorder(config.ManagementURL.String())
 	r.GetFullStatus()
 
-	connectClient := internal.NewConnectClient(ctx, config, r, false)
+	connectClient := internal.NewConnectClient(ctx, config, r)
 	SetupDebugHandler(ctx, config, r, connectClient, "")
 
-	return connectClient.Run(nil)
+	return connectClient.Run(nil, util.FindFirstLogPath(logFiles))
 }
 
 func runInDaemonMode(ctx context.Context, cmd *cobra.Command, pm *profilemanager.ProfileManager, activeProf *profilemanager.Profile, profileSwitched bool) error {
@@ -216,6 +216,7 @@ func runInDaemonMode(ctx context.Context, cmd *cobra.Command, pm *profilemanager
 
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
+		//nolint
 		return fmt.Errorf("failed to connect to daemon error: %v\n"+
 			"If the daemon is not running please run: "+
 			"\nnetbird service install \nnetbird service start\n", err)
@@ -429,6 +430,10 @@ func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, pro
 		req.BlockInbound = &blockInbound
 	}
 
+	if cmd.Flag(disableIPv6Flag).Changed {
+		req.DisableIpv6 = &disableIPv6
+	}
+
 	if cmd.Flag(enableLazyConnectionFlag).Changed {
 		req.LazyConnectionEnabled = &lazyConnEnabled
 	}
@@ -546,6 +551,10 @@ func setupConfig(customDNSAddressConverted []byte, cmd *cobra.Command, configFil
 		ic.BlockInbound = &blockInbound
 	}
 
+	if cmd.Flag(disableIPv6Flag).Changed {
+		ic.DisableIPv6 = &disableIPv6
+	}
+
 	if cmd.Flag(enableLazyConnectionFlag).Changed {
 		ic.LazyConnectionEnabled = &lazyConnEnabled
 	}
@@ -660,6 +669,10 @@ func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte
 		loginRequest.BlockInbound = &blockInbound
 	}
 
+	if cmd.Flag(disableIPv6Flag).Changed {
+		loginRequest.DisableIpv6 = &disableIPv6
+	}
+
 	if cmd.Flag(enableLazyConnectionFlag).Changed {
 		loginRequest.LazyConnectionEnabled = &lazyConnEnabled
 	}

client/cmd/update_supported.go

@@ -11,7 +11,7 @@ import (
 	log "github.com/sirupsen/logrus"
 	"github.com/spf13/cobra"
 
-	"github.com/netbirdio/netbird/client/internal/updatemanager/installer"
+	"github.com/netbirdio/netbird/client/internal/updater/installer"
 	"github.com/netbirdio/netbird/util"
 )
 

client/embed/embed.go

@@ -14,12 +14,16 @@ import (
 	"github.com/sirupsen/logrus"
 	wgnetstack "golang.zx2c4.com/wireguard/tun/netstack"
 
+	"github.com/netbirdio/netbird/client/iface"
 	"github.com/netbirdio/netbird/client/iface/netstack"
 	"github.com/netbirdio/netbird/client/internal"
+	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/peer"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	sshcommon "github.com/netbirdio/netbird/client/ssh"
 	"github.com/netbirdio/netbird/client/system"
+	"github.com/netbirdio/netbird/shared/management/domain"
+	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
 )
 
 var (
@@ -29,6 +33,14 @@ var (
 	ErrConfigNotInitialized = errors.New("config not initialized")
 )
 
+// PeerConnStatus is a peer's connection status.
+type PeerConnStatus = peer.ConnStatus
+
+const (
+	// PeerStatusConnected indicates the peer is in connected state.
+	PeerStatusConnected = peer.StatusConnected
+)
+
 // Client manages a netbird embedded client instance.
 type Client struct {
 	deviceName string
@@ -38,6 +50,7 @@ type Client struct {
 	setupKey   string
 	jwtToken   string
 	connect    *internal.ConnectClient
+	recorder   *peer.Status
 }
 
 // Options configures a new Client.
@@ -66,6 +79,20 @@ type Options struct {
 	StatePath string
 	// DisableClientRoutes disables the client routes
 	DisableClientRoutes bool
+	// DisableIPv6 disables IPv6 overlay addressing
+	DisableIPv6 bool
+	// BlockInbound blocks all inbound connections from peers
+	BlockInbound bool
+	// WireguardPort is the port for the WireGuard interface. Use 0 for a random port.
+	WireguardPort *int
+	// MTU is the MTU for the WireGuard interface.
+	// Valid values are in the range 576..8192 bytes.
+	// If non-nil, this value overrides any value stored in the config file.
+	// If nil, the existing config MTU (if non-zero) is preserved; otherwise it defaults to 1280.
+	// Set to a higher value (e.g. 1400) if carrying QUIC or other protocols that require larger datagrams.
+	MTU *uint16
+	// DNSLabels defines additional DNS labels configured in the peer.
+	DNSLabels []string
 }
 
 // validateCredentials checks that exactly one credential type is provided
@@ -97,6 +124,12 @@ func New(opts Options) (*Client, error) {
 		return nil, err
 	}
 
+	if opts.MTU != nil {
+		if err := iface.ValidateMTU(*opts.MTU); err != nil {
+			return nil, fmt.Errorf("invalid MTU: %w", err)
+		}
+	}
+
 	if opts.LogOutput != nil {
 		logrus.SetOutput(opts.LogOutput)
 	}
@@ -125,15 +158,25 @@ func New(opts Options) (*Client, error) {
 		}
 	}
 
+	var err error
+	var parsedLabels domain.List
+	if parsedLabels, err = domain.FromStringList(opts.DNSLabels); err != nil {
+		return nil, fmt.Errorf("invalid dns labels: %w", err)
+	}
+
 	t := true
 	var config *profilemanager.Config
-	var err error
 	input := profilemanager.ConfigInput{
 		ConfigPath:          opts.ConfigPath,
 		ManagementURL:       opts.ManagementURL,
 		PreSharedKey:        &opts.PreSharedKey,
 		DisableServerRoutes: &t,
 		DisableClientRoutes: &opts.DisableClientRoutes,
+		DisableIPv6:         &opts.DisableIPv6,
+		BlockInbound:        &opts.BlockInbound,
+		WireguardPort:       opts.WireguardPort,
+		MTU:                 opts.MTU,
+		DNSLabels:           parsedLabels,
 	}
 	if opts.ConfigPath != "" {
 		config, err = profilemanager.UpdateOrCreateConfig(input)
@@ -153,6 +196,7 @@ func New(opts Options) (*Client, error) {
 		setupKey:   opts.SetupKey,
 		jwtToken:   opts.JWTToken,
 		config:     config,
+		recorder:   peer.NewRecorder(config.ManagementURL.String()),
 	}, nil
 }
 
@@ -161,26 +205,38 @@ func New(opts Options) (*Client, error) {
 func (c *Client) Start(startCtx context.Context) error {
 	c.mu.Lock()
 	defer c.mu.Unlock()
-	if c.cancel != nil {
+	if c.connect != nil {
 		return ErrClientAlreadyStarted
 	}
 
-	ctx := internal.CtxInitState(context.Background())
+	ctx, cancel := context.WithCancel(internal.CtxInitState(context.Background()))
+	defer func() {
+		if c.connect == nil {
+			cancel()
+		}
+	}()
+
 	// nolint:staticcheck
 	ctx = context.WithValue(ctx, system.DeviceNameCtxKey, c.deviceName)
-	if err := internal.Login(ctx, c.config, c.setupKey, c.jwtToken); err != nil {
+
+	authClient, err := auth.NewAuth(ctx, c.config.PrivateKey, c.config.ManagementURL, c.config)
+	if err != nil {
+		return fmt.Errorf("create auth client: %w", err)
+	}
+	defer authClient.Close()
+
+	if err, _ := authClient.Login(ctx, c.setupKey, c.jwtToken); err != nil {
 		return fmt.Errorf("login: %w", err)
 	}
-
-	recorder := peer.NewRecorder(c.config.ManagementURL.String())
-	client := internal.NewConnectClient(ctx, c.config, recorder, false)
+	client := internal.NewConnectClient(ctx, c.config, c.recorder)
+	client.SetSyncResponsePersistence(true)
 
 	// either startup error (permanent backoff err) or nil err (successful engine up)
 	// TODO: make after-startup backoff err available
 	run := make(chan struct{})
 	clientErr := make(chan error, 1)
 	go func() {
-		if err := client.Run(run); err != nil {
+		if err := client.Run(run, ""); err != nil {
 			clientErr <- err
 		}
 	}()
@@ -197,6 +253,7 @@ func (c *Client) Start(startCtx context.Context) error {
 	}
 
 	c.connect = client
+	c.cancel = cancel
 
 	return nil
 }
@@ -211,17 +268,23 @@ func (c *Client) Stop(ctx context.Context) error {
 		return ErrClientNotStarted
 	}
 
+	if c.cancel != nil {
+		c.cancel()
+		c.cancel = nil
+	}
+
 	done := make(chan error, 1)
+	connect := c.connect
 	go func() {
-		done <- c.connect.Stop()
+		done <- connect.Stop()
 	}()
 
 	select {
 	case <-ctx.Done():
-		c.cancel = nil
+		c.connect = nil
 		return ctx.Err()
 	case err := <-done:
-		c.cancel = nil
+		c.connect = nil
 		if err != nil {
 			return fmt.Errorf("stop: %w", err)
 		}
@@ -315,6 +378,57 @@ func (c *Client) NewHTTPClient() *http.Client {
 	}
 }
 
+// Status returns the current status of the client.
+func (c *Client) Status() (peer.FullStatus, error) {
+	c.mu.Lock()
+	connect := c.connect
+	c.mu.Unlock()
+
+	if connect != nil {
+		engine := connect.Engine()
+		if engine != nil {
+			_ = engine.RunHealthProbes(false)
+		}
+	}
+
+	return c.recorder.GetFullStatus(), nil
+}
+
+// GetLatestSyncResponse returns the latest sync response from the management server.
+func (c *Client) GetLatestSyncResponse() (*mgmProto.SyncResponse, error) {
+	engine, err := c.getEngine()
+	if err != nil {
+		return nil, err
+	}
+
+	syncResp, err := engine.GetLatestSyncResponse()
+	if err != nil {
+		return nil, fmt.Errorf("get sync response: %w", err)
+	}
+
+	return syncResp, nil
+}
+
+// SetLogLevel sets the logging level for the client and its components.
+func (c *Client) SetLogLevel(levelStr string) error {
+	level, err := logrus.ParseLevel(levelStr)
+	if err != nil {
+		return fmt.Errorf("parse log level: %w", err)
+	}
+
+	logrus.SetLevel(level)
+
+	c.mu.Lock()
+	connect := c.connect
+	c.mu.Unlock()
+
+	if connect != nil {
+		connect.SetLogLevel(level)
+	}
+
+	return nil
+}
+
 // VerifySSHHostKey verifies an SSH host key against stored peer keys.
 // Returns nil if the key matches, ErrPeerNotFound if peer is not in network,
 // ErrNoStoredKey if peer has no stored key, or an error for verification failures.

client/firewall/iptables/acl_linux.go

@@ -36,6 +36,7 @@ type aclManager struct {
 	entries         aclEntries
 	optionalEntries map[string][]entry
 	ipsetStore      *ipsetStore
+	v6              bool
 
 	stateManager *statemanager.Manager
 }
@@ -47,6 +48,7 @@ func newAclManager(iptablesClient *iptables.IPTables, wgIface iFaceMapper) (*acl
 		entries:         make(map[string][][]string),
 		optionalEntries: make(map[string][]entry),
 		ipsetStore:      newIpsetStore(),
+		v6:              iptablesClient.Proto() == iptables.ProtocolIPv6,
 	}, nil
 }
 
@@ -81,7 +83,11 @@ func (m *aclManager) AddPeerFiltering(
 	chain := chainNameInputRules
 
 	ipsetName = transformIPsetName(ipsetName, sPort, dPort, action)
-	specs := filterRuleSpecs(ip, string(protocol), sPort, dPort, action, ipsetName)
+	if m.v6 && ipsetName != "" {
+		ipsetName += "-v6"
+	}
+	proto := protoForFamily(protocol, m.v6)
+	specs := filterRuleSpecs(ip, proto, sPort, dPort, action, ipsetName)
 
 	mangleSpecs := slices.Clone(specs)
 	mangleSpecs = append(mangleSpecs,
@@ -105,6 +111,7 @@ func (m *aclManager) AddPeerFiltering(
 				ip:        ip.String(),
 				chain:     chain,
 				specs:     specs,
+				v6:        m.v6,
 			}}, nil
 		}
 
@@ -157,6 +164,7 @@ func (m *aclManager) AddPeerFiltering(
 		ipsetName:   ipsetName,
 		ip:          ip.String(),
 		chain:       chain,
+		v6:          m.v6,
 	}
 
 	m.updateState()
@@ -376,8 +384,13 @@ func (m *aclManager) updateState() {
 	currentState.Lock()
 	defer currentState.Unlock()
 
-	currentState.ACLEntries = m.entries
-	currentState.ACLIPsetStore = m.ipsetStore
+	if m.v6 {
+		currentState.ACLEntries6 = m.entries
+		currentState.ACLIPsetStore6 = m.ipsetStore
+	} else {
+		currentState.ACLEntries = m.entries
+		currentState.ACLIPsetStore = m.ipsetStore
+	}
 
 	if err := m.stateManager.UpdateState(currentState); err != nil {
 		log.Errorf("failed to update state: %v", err)
@@ -385,12 +398,18 @@ func (m *aclManager) updateState() {
 }
 
 // filterRuleSpecs returns the specs of a filtering rule
-func filterRuleSpecs(ip net.IP, protocol string, sPort, dPort *firewall.Port, action firewall.Action, ipsetName string) (specs []string) {
-	matchByIP := true
-	// don't use IP matching if IP is 0.0.0.0
-	if ip.IsUnspecified() {
-		matchByIP = false
+// protoForFamily translates ICMP to ICMPv6 for ip6tables.
+// ip6tables requires "ipv6-icmp" (or "icmpv6") instead of "icmp".
+func protoForFamily(protocol firewall.Protocol, v6 bool) string {
+	if v6 && protocol == firewall.ProtocolICMP {
+		return "ipv6-icmp"
 	}
+	return string(protocol)
+}
+
+func filterRuleSpecs(ip net.IP, protocol string, sPort, dPort *firewall.Port, action firewall.Action, ipsetName string) (specs []string) {
+	// don't use IP matching if IP is 0.0.0.0
+	matchByIP := !ip.IsUnspecified()
 
 	if matchByIP {
 		if ipsetName != "" {
@@ -440,6 +459,9 @@ func (m *aclManager) createIPSet(name string) error {
 	opts := ipset.CreateOptions{
 		Replace: true,
 	}
+	if m.v6 {
+		opts.Family = ipset.FamilyIPV6
+	}
 
 	if err := ipset.Create(name, ipset.TypeHashNet, opts); err != nil {
 		return fmt.Errorf("create ipset %s: %w", name, err)

client/firewall/iptables/manager_linux.go

@@ -23,9 +23,15 @@ type Manager struct {
 
 	wgIface iFaceMapper
 
-	ipv4Client *iptables.IPTables
-	aclMgr     *aclManager
-	router     *router
+	ipv4Client   *iptables.IPTables
+	aclMgr       *aclManager
+	router       *router
+	rawSupported bool
+
+	// IPv6 counterparts, nil when no v6 overlay
+	ipv6Client *iptables.IPTables
+	aclMgr6    *aclManager
+	router6    *router
 }
 
 // iFaceMapper defines subset methods of interface required for manager
@@ -57,9 +63,39 @@ func Create(wgIface iFaceMapper, mtu uint16) (*Manager, error) {
 		return nil, fmt.Errorf("create acl manager: %w", err)
 	}
 
+	if wgIface.Address().HasIPv6() {
+		if err := m.createIPv6Components(wgIface, mtu); err != nil {
+			return nil, fmt.Errorf("create IPv6 firewall: %w", err)
+		}
+	}
+
 	return m, nil
 }
 
+func (m *Manager) createIPv6Components(wgIface iFaceMapper, mtu uint16) error {
+	ip6Client, err := iptables.NewWithProtocol(iptables.ProtocolIPv6)
+	if err != nil {
+		return fmt.Errorf("init ip6tables: %w", err)
+	}
+	m.ipv6Client = ip6Client
+
+	m.router6, err = newRouter(ip6Client, wgIface, mtu)
+	if err != nil {
+		return fmt.Errorf("create v6 router: %w", err)
+	}
+
+	m.aclMgr6, err = newAclManager(ip6Client, wgIface)
+	if err != nil {
+		return fmt.Errorf("create v6 acl manager: %w", err)
+	}
+
+	return nil
+}
+
+func (m *Manager) hasIPv6() bool {
+	return m.ipv6Client != nil
+}
+
 func (m *Manager) Init(stateManager *statemanager.Manager) error {
 	state := &ShutdownState{
 		InterfaceState: &InterfaceState{
@@ -83,6 +119,19 @@ func (m *Manager) Init(stateManager *statemanager.Manager) error {
 		return fmt.Errorf("acl manager init: %w", err)
 	}
 
+	if m.hasIPv6() {
+		if err := m.router6.init(stateManager); err != nil {
+			return fmt.Errorf("v6 router init: %w", err)
+		}
+		if err := m.aclMgr6.init(stateManager); err != nil {
+			return fmt.Errorf("v6 acl manager init: %w", err)
+		}
+	}
+
+	if err := m.initNoTrackChain(); err != nil {
+		log.Warnf("raw table not available, notrack rules will be disabled: %v", err)
+	}
+
 	// persist early to ensure cleanup of chains
 	go func() {
 		if err := stateManager.PersistState(context.Background()); err != nil {
@@ -108,7 +157,13 @@ func (m *Manager) AddPeerFiltering(
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.aclMgr.AddPeerFiltering(id, ip, proto, sPort, dPort, action, ipsetName)
+	if ip.To4() != nil {
+		return m.aclMgr.AddPeerFiltering(id, ip, proto, sPort, dPort, action, ipsetName)
+	}
+	if !m.hasIPv6() {
+		return nil, fmt.Errorf("IPv6 not initialized, cannot add rule for %s", ip)
+	}
+	return m.aclMgr6.AddPeerFiltering(id, ip, proto, sPort, dPort, action, ipsetName)
 }
 
 func (m *Manager) AddRouteFiltering(
@@ -122,25 +177,48 @@ func (m *Manager) AddRouteFiltering(
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	if destination.IsPrefix() && !destination.Prefix.Addr().Is4() {
-		return nil, fmt.Errorf("unsupported IP version: %s", destination.Prefix.Addr().String())
+	if isIPv6RouteRule(sources, destination) {
+		if !m.hasIPv6() {
+			return nil, fmt.Errorf("IPv6 not initialized, cannot add route rule")
+		}
+		return m.router6.AddRouteFiltering(id, sources, destination, proto, sPort, dPort, action)
 	}
 
 	return m.router.AddRouteFiltering(id, sources, destination, proto, sPort, dPort, action)
 }
 
+func isIPv6RouteRule(sources []netip.Prefix, destination firewall.Network) bool {
+	if destination.IsPrefix() {
+		return destination.Prefix.Addr().Is6()
+	}
+	return len(sources) > 0 && sources[0].Addr().Is6()
+}
+
 // DeletePeerRule from the firewall by rule definition
 func (m *Manager) DeletePeerRule(rule firewall.Rule) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && isIPv6IptRule(rule) {
+		return m.aclMgr6.DeletePeerRule(rule)
+	}
 	return m.aclMgr.DeletePeerRule(rule)
 }
 
+func isIPv6IptRule(rule firewall.Rule) bool {
+	r, ok := rule.(*Rule)
+	return ok && r.v6
+}
+
+// DeleteRouteRule deletes a routing rule.
+// Route rules are keyed by content hash. Check v4 first, try v6 if not found.
 func (m *Manager) DeleteRouteRule(rule firewall.Rule) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && !m.router.hasRule(rule.ID()) {
+		return m.router6.DeleteRouteRule(rule)
+	}
 	return m.router.DeleteRouteRule(rule)
 }
 
@@ -156,18 +234,63 @@ func (m *Manager) AddNatRule(pair firewall.RouterPair) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.AddNatRule(pair)
+	if pair.Destination.IsPrefix() && pair.Destination.Prefix.Addr().Is6() {
+		if !m.hasIPv6() {
+			return fmt.Errorf("IPv6 not initialized, cannot add NAT rule")
+		}
+		return m.router6.AddNatRule(pair)
+	}
+
+	if err := m.router.AddNatRule(pair); err != nil {
+		return err
+	}
+
+	// Dynamic routes need NAT in both tables
+	if m.hasIPv6() && pair.Destination.IsSet() {
+		v6Pair := pair
+		v6Pair.Source = firewall.Network{Prefix: netip.PrefixFrom(netip.IPv6Unspecified(), 0)}
+		if err := m.router6.AddNatRule(v6Pair); err != nil {
+			return fmt.Errorf("add v6 NAT rule: %w", err)
+		}
+	}
+
+	return nil
 }
 
 func (m *Manager) RemoveNatRule(pair firewall.RouterPair) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.RemoveNatRule(pair)
+	if pair.Destination.IsPrefix() && pair.Destination.Prefix.Addr().Is6() {
+		if !m.hasIPv6() {
+			return nil
+		}
+		return m.router6.RemoveNatRule(pair)
+	}
+
+	if err := m.router.RemoveNatRule(pair); err != nil {
+		return err
+	}
+
+	if m.hasIPv6() && pair.Destination.IsSet() {
+		v6Pair := pair
+		v6Pair.Source = firewall.Network{Prefix: netip.PrefixFrom(netip.IPv6Unspecified(), 0)}
+		if err := m.router6.RemoveNatRule(v6Pair); err != nil {
+			return fmt.Errorf("remove v6 NAT rule: %w", err)
+		}
+	}
+
+	return nil
 }
 
 func (m *Manager) SetLegacyManagement(isLegacy bool) error {
-	return firewall.SetLegacyManagement(m.router, isLegacy)
+	if err := firewall.SetLegacyManagement(m.router, isLegacy); err != nil {
+		return err
+	}
+	if m.hasIPv6() {
+		return firewall.SetLegacyManagement(m.router6, isLegacy)
+	}
+	return nil
 }
 
 // Reset firewall to the default state
@@ -177,6 +300,19 @@ func (m *Manager) Close(stateManager *statemanager.Manager) error {
 
 	var merr *multierror.Error
 
+	if err := m.cleanupNoTrackChain(); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("cleanup notrack chain: %w", err))
+	}
+
+	if m.hasIPv6() {
+		if err := m.aclMgr6.Reset(); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("reset v6 acl manager: %w", err))
+		}
+		if err := m.router6.Reset(); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("reset v6 router: %w", err))
+		}
+	}
+
 	if err := m.aclMgr.Reset(); err != nil {
 		merr = multierror.Append(merr, fmt.Errorf("reset acl manager: %w", err))
 	}
@@ -200,19 +336,16 @@ func (m *Manager) AllowNetbird() error {
 		return nil
 	}
 
-	_, err := m.AddPeerFiltering(
-		nil,
-		net.IP{0, 0, 0, 0},
-		firewall.ProtocolALL,
-		nil,
-		nil,
-		firewall.ActionAccept,
-		"",
-	)
-	if err != nil {
-		return fmt.Errorf("allow netbird interface traffic: %w", err)
+	var merr *multierror.Error
+	if _, err := m.aclMgr.AddPeerFiltering(nil, net.IP{0, 0, 0, 0}, firewall.ProtocolALL, nil, nil, firewall.ActionAccept, ""); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("allow netbird interface traffic: %w", err))
 	}
-	return nil
+	if m.hasIPv6() {
+		if _, err := m.aclMgr6.AddPeerFiltering(nil, net.IPv6zero, firewall.ProtocolALL, nil, nil, firewall.ActionAccept, ""); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("allow v6 netbird interface traffic: %w", err))
+		}
+	}
+	return nberrors.FormatErrorOrNil(merr)
 }
 
 // Flush doesn't need to be implemented for this manager
@@ -242,6 +375,9 @@ func (m *Manager) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error)
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && rule.TranslatedAddress.Is6() {
+		return m.router6.AddDNATRule(rule)
+	}
 	return m.router.AddDNATRule(rule)
 }
 
@@ -250,6 +386,9 @@ func (m *Manager) DeleteDNATRule(rule firewall.Rule) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && !m.router.hasRule(rule.ID()+dnatSuffix) {
+		return m.router6.DeleteDNATRule(rule)
+	}
 	return m.router.DeleteDNATRule(rule)
 }
 
@@ -258,7 +397,26 @@ func (m *Manager) UpdateSet(set firewall.Set, prefixes []netip.Prefix) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.UpdateSet(set, prefixes)
+	var v4Prefixes, v6Prefixes []netip.Prefix
+	for _, p := range prefixes {
+		if p.Addr().Is6() {
+			v6Prefixes = append(v6Prefixes, p)
+		} else {
+			v4Prefixes = append(v4Prefixes, p)
+		}
+	}
+
+	if err := m.router.UpdateSet(set, v4Prefixes); err != nil {
+		return err
+	}
+
+	if m.hasIPv6() && len(v6Prefixes) > 0 {
+		if err := m.router6.UpdateSet(set, v6Prefixes); err != nil {
+			return fmt.Errorf("update v6 set: %w", err)
+		}
+	}
+
+	return nil
 }
 
 // AddInboundDNAT adds an inbound DNAT rule redirecting traffic from NetBird peers to local services.
@@ -266,6 +424,9 @@ func (m *Manager) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protoco
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && localAddr.Is6() {
+		return m.router6.AddInboundDNAT(localAddr, protocol, sourcePort, targetPort)
+	}
 	return m.router.AddInboundDNAT(localAddr, protocol, sourcePort, targetPort)
 }
 
@@ -274,9 +435,140 @@ func (m *Manager) RemoveInboundDNAT(localAddr netip.Addr, protocol firewall.Prot
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && localAddr.Is6() {
+		return m.router6.RemoveInboundDNAT(localAddr, protocol, sourcePort, targetPort)
+	}
 	return m.router.RemoveInboundDNAT(localAddr, protocol, sourcePort, targetPort)
 }
 
+const (
+	chainNameRaw = "NETBIRD-RAW"
+	chainOUTPUT  = "OUTPUT"
+	tableRaw     = "raw"
+)
+
+// SetupEBPFProxyNoTrack creates notrack rules for eBPF proxy loopback traffic.
+// This prevents conntrack from tracking WireGuard proxy traffic on loopback, which
+// can interfere with MASQUERADE rules (e.g., from container runtimes like Podman/netavark).
+//
+// Traffic flows that need NOTRACK:
+//
+//  1. Egress: WireGuard -> fake endpoint (before eBPF rewrite)
+//     src=127.0.0.1:wgPort -> dst=127.0.0.1:fakePort
+//     Matched by: sport=wgPort
+//
+//  2. Egress: Proxy -> WireGuard (via raw socket)
+//     src=127.0.0.1:fakePort -> dst=127.0.0.1:wgPort
+//     Matched by: dport=wgPort
+//
+//  3. Ingress: Packets to WireGuard
+//     dst=127.0.0.1:wgPort
+//     Matched by: dport=wgPort
+//
+//  4. Ingress: Packets to proxy (after eBPF rewrite)
+//     dst=127.0.0.1:proxyPort
+//     Matched by: dport=proxyPort
+//
+// Rules are cleaned up when the firewall manager is closed.
+func (m *Manager) SetupEBPFProxyNoTrack(proxyPort, wgPort uint16) error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	if !m.rawSupported {
+		return fmt.Errorf("raw table not available")
+	}
+
+	wgPortStr := fmt.Sprintf("%d", wgPort)
+	proxyPortStr := fmt.Sprintf("%d", proxyPort)
+
+	// Egress rules: match outgoing loopback UDP packets
+	outputRuleSport := []string{"-o", "lo", "-s", "127.0.0.1", "-d", "127.0.0.1", "-p", "udp", "--sport", wgPortStr, "-j", "NOTRACK"}
+	if err := m.ipv4Client.AppendUnique(tableRaw, chainNameRaw, outputRuleSport...); err != nil {
+		return fmt.Errorf("add output sport notrack rule: %w", err)
+	}
+
+	outputRuleDport := []string{"-o", "lo", "-s", "127.0.0.1", "-d", "127.0.0.1", "-p", "udp", "--dport", wgPortStr, "-j", "NOTRACK"}
+	if err := m.ipv4Client.AppendUnique(tableRaw, chainNameRaw, outputRuleDport...); err != nil {
+		return fmt.Errorf("add output dport notrack rule: %w", err)
+	}
+
+	// Ingress rules: match incoming loopback UDP packets
+	preroutingRuleWg := []string{"-i", "lo", "-s", "127.0.0.1", "-d", "127.0.0.1", "-p", "udp", "--dport", wgPortStr, "-j", "NOTRACK"}
+	if err := m.ipv4Client.AppendUnique(tableRaw, chainNameRaw, preroutingRuleWg...); err != nil {
+		return fmt.Errorf("add prerouting wg notrack rule: %w", err)
+	}
+
+	preroutingRuleProxy := []string{"-i", "lo", "-s", "127.0.0.1", "-d", "127.0.0.1", "-p", "udp", "--dport", proxyPortStr, "-j", "NOTRACK"}
+	if err := m.ipv4Client.AppendUnique(tableRaw, chainNameRaw, preroutingRuleProxy...); err != nil {
+		return fmt.Errorf("add prerouting proxy notrack rule: %w", err)
+	}
+
+	log.Debugf("set up ebpf proxy notrack rules for ports %d,%d", proxyPort, wgPort)
+	return nil
+}
+
+func (m *Manager) initNoTrackChain() error {
+	if err := m.cleanupNoTrackChain(); err != nil {
+		log.Debugf("cleanup notrack chain: %v", err)
+	}
+
+	if err := m.ipv4Client.NewChain(tableRaw, chainNameRaw); err != nil {
+		return fmt.Errorf("create chain: %w", err)
+	}
+
+	jumpRule := []string{"-j", chainNameRaw}
+
+	if err := m.ipv4Client.InsertUnique(tableRaw, chainOUTPUT, 1, jumpRule...); err != nil {
+		if delErr := m.ipv4Client.DeleteChain(tableRaw, chainNameRaw); delErr != nil {
+			log.Debugf("delete orphan chain: %v", delErr)
+		}
+		return fmt.Errorf("add output jump rule: %w", err)
+	}
+
+	if err := m.ipv4Client.InsertUnique(tableRaw, chainPREROUTING, 1, jumpRule...); err != nil {
+		if delErr := m.ipv4Client.DeleteIfExists(tableRaw, chainOUTPUT, jumpRule...); delErr != nil {
+			log.Debugf("delete output jump rule: %v", delErr)
+		}
+		if delErr := m.ipv4Client.DeleteChain(tableRaw, chainNameRaw); delErr != nil {
+			log.Debugf("delete orphan chain: %v", delErr)
+		}
+		return fmt.Errorf("add prerouting jump rule: %w", err)
+	}
+
+	m.rawSupported = true
+	return nil
+}
+
+func (m *Manager) cleanupNoTrackChain() error {
+	exists, err := m.ipv4Client.ChainExists(tableRaw, chainNameRaw)
+	if err != nil {
+		if !m.rawSupported {
+			return nil
+		}
+		return fmt.Errorf("check chain exists: %w", err)
+	}
+	if !exists {
+		return nil
+	}
+
+	jumpRule := []string{"-j", chainNameRaw}
+
+	if err := m.ipv4Client.DeleteIfExists(tableRaw, chainOUTPUT, jumpRule...); err != nil {
+		return fmt.Errorf("remove output jump rule: %w", err)
+	}
+
+	if err := m.ipv4Client.DeleteIfExists(tableRaw, chainPREROUTING, jumpRule...); err != nil {
+		return fmt.Errorf("remove prerouting jump rule: %w", err)
+	}
+
+	if err := m.ipv4Client.ClearAndDeleteChain(tableRaw, chainNameRaw); err != nil {
+		return fmt.Errorf("clear and delete chain: %w", err)
+	}
+
+	m.rawSupported = false
+	return nil
+}
+
 func getConntrackEstablished() []string {
 	return []string{"-m", "conntrack", "--ctstate", "RELATED,ESTABLISHED", "-j", "ACCEPT"}
 }

client/firewall/iptables/manager_linux_test.go

@@ -161,7 +161,7 @@ func TestIptablesManagerDenyRules(t *testing.T) {
 			t.Logf("  [%d] %s", i, rule)
 		}
 
-		var denyRuleIndex, acceptRuleIndex int = -1, -1
+		var denyRuleIndex, acceptRuleIndex = -1, -1
 		for i, rule := range rules {
 			if strings.Contains(rule, "DROP") {
 				t.Logf("Found DROP rule at index %d: %s", i, rule)

client/firewall/iptables/router_linux.go

@@ -52,8 +52,10 @@ const (
 	snatSuffix = "_snat"
 	fwdSuffix  = "_fwd"
 
-	// ipTCPHeaderMinSize represents minimum IP (20) + TCP (20) header size for MSS calculation
-	ipTCPHeaderMinSize = 40
+	// ipv4TCPHeaderSize is the minimum IPv4 (20) + TCP (20) header size for MSS calculation.
+	ipv4TCPHeaderSize = 40
+	// ipv6TCPHeaderSize is the minimum IPv6 (40) + TCP (20) header size for MSS calculation.
+	ipv6TCPHeaderSize = 60
 )
 
 type ruleInfo struct {
@@ -84,6 +86,7 @@ type router struct {
 	wgIface          iFaceMapper
 	legacyManagement bool
 	mtu              uint16
+	v6               bool
 
 	stateManager *statemanager.Manager
 	ipFwdState   *ipfwdstate.IPForwardingState
@@ -95,6 +98,7 @@ func newRouter(iptablesClient *iptables.IPTables, wgIface iFaceMapper, mtu uint1
 		rules:          make(map[string][]string),
 		wgIface:        wgIface,
 		mtu:            mtu,
+		v6:             iptablesClient.Proto() == iptables.ProtocolIPv6,
 		ipFwdState:     ipfwdstate.NewIPForwardingState(),
 	}
 
@@ -184,6 +188,11 @@ func (r *router) AddRouteFiltering(
 	return ruleKey, nil
 }
 
+func (r *router) hasRule(id string) bool {
+	_, ok := r.rules[id]
+	return ok
+}
+
 func (r *router) DeleteRouteRule(rule firewall.Rule) error {
 	ruleKey := rule.ID()
 
@@ -423,6 +432,12 @@ func (r *router) createContainers() error {
 		{chainRTRDR, tableNat},
 		{chainRTMSSCLAMP, tableMangle},
 	} {
+		// Fallback: clear chains that survived an unclean shutdown.
+		if ok, _ := r.iptablesClient.ChainExists(chainInfo.table, chainInfo.chain); ok {
+			if err := r.iptablesClient.ClearAndDeleteChain(chainInfo.table, chainInfo.chain); err != nil {
+				log.Warnf("clear stale chain %s in %s: %v", chainInfo.chain, chainInfo.table, err)
+			}
+		}
 		if err := r.iptablesClient.NewChain(chainInfo.table, chainInfo.chain); err != nil {
 			return fmt.Errorf("create chain %s in table %s: %w", chainInfo.chain, chainInfo.table, err)
 		}
@@ -529,9 +544,12 @@ func (r *router) addPostroutingRules() error {
 }
 
 // addMSSClampingRules adds MSS clamping rules to prevent fragmentation for forwarded traffic.
-// TODO: Add IPv6 support
 func (r *router) addMSSClampingRules() error {
-	mss := r.mtu - ipTCPHeaderMinSize
+	overhead := uint16(ipv4TCPHeaderSize)
+	if r.v6 {
+		overhead = ipv6TCPHeaderSize
+	}
+	mss := r.mtu - overhead
 
 	// Add jump rule from FORWARD chain in mangle table to our custom chain
 	jumpRule := []string{
@@ -716,8 +734,13 @@ func (r *router) updateState() {
 	currentState.Lock()
 	defer currentState.Unlock()
 
-	currentState.RouteRules = r.rules
-	currentState.RouteIPsetCounter = r.ipsetCounter
+	if r.v6 {
+		currentState.RouteRules6 = r.rules
+		currentState.RouteIPsetCounter6 = r.ipsetCounter
+	} else {
+		currentState.RouteRules = r.rules
+		currentState.RouteIPsetCounter = r.ipsetCounter
+	}
 
 	if err := r.stateManager.UpdateState(currentState); err != nil {
 		log.Errorf("failed to update state: %v", err)
@@ -845,7 +868,7 @@ func (r *router) DeleteDNATRule(rule firewall.Rule) error {
 	}
 
 	if fwdRule, exists := r.rules[ruleKey+fwdSuffix]; exists {
-		if err := r.iptablesClient.Delete(tableFilter, chainRTFWDIN, fwdRule...); err != nil {
+		if err := r.iptablesClient.Delete(tableFilter, chainRTFWDOUT, fwdRule...); err != nil {
 			merr = multierror.Append(merr, fmt.Errorf("delete forward rule: %w", err))
 		}
 		delete(r.rules, ruleKey+fwdSuffix)
@@ -872,7 +895,7 @@ func (r *router) genRouteRuleSpec(params routeFilteringRuleParams, sources []net
 	rule = append(rule, destExp...)
 
 	if params.Proto != firewall.ProtocolALL {
-		rule = append(rule, "-p", strings.ToLower(string(params.Proto)))
+		rule = append(rule, "-p", strings.ToLower(protoForFamily(params.Proto, r.v6)))
 		rule = append(rule, applyPort("--sport", params.SPort)...)
 		rule = append(rule, applyPort("--dport", params.DPort)...)
 	}
@@ -889,11 +912,12 @@ func (r *router) applyNetwork(flag string, network firewall.Network, prefixes []
 	}
 
 	if network.IsSet() {
-		if _, err := r.ipsetCounter.Increment(network.Set.HashedName(), prefixes); err != nil {
+		name := r.ipsetName(network.Set.HashedName())
+		if _, err := r.ipsetCounter.Increment(name, prefixes); err != nil {
 			return nil, fmt.Errorf("create or get ipset: %w", err)
 		}
 
-		return []string{"-m", "set", matchSet, network.Set.HashedName(), direction}, nil
+		return []string{"-m", "set", matchSet, name, direction}, nil
 	}
 	if network.IsPrefix() {
 		return []string{flag, network.Prefix.String()}, nil
@@ -904,19 +928,15 @@ func (r *router) applyNetwork(flag string, network firewall.Network, prefixes []
 }
 
 func (r *router) UpdateSet(set firewall.Set, prefixes []netip.Prefix) error {
+	name := r.ipsetName(set.HashedName())
 	var merr *multierror.Error
 	for _, prefix := range prefixes {
-		// TODO: Implement IPv6 support
-		if prefix.Addr().Is6() {
-			log.Tracef("skipping IPv6 prefix %s: IPv6 support not yet implemented", prefix)
-			continue
-		}
-		if err := r.addPrefixToIPSet(set.HashedName(), prefix); err != nil {
+		if err := r.addPrefixToIPSet(name, prefix); err != nil {
 			merr = multierror.Append(merr, fmt.Errorf("add prefix to ipset: %w", err))
 		}
 	}
 	if merr == nil {
-		log.Debugf("updated set %s with prefixes %v", set.HashedName(), prefixes)
+		log.Debugf("updated set %s with prefixes %v", name, prefixes)
 	}
 
 	return nberrors.FormatErrorOrNil(merr)
@@ -932,7 +952,7 @@ func (r *router) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol
 
 	dnatRule := []string{
 		"-i", r.wgIface.Name(),
-		"-p", strings.ToLower(string(protocol)),
+		"-p", strings.ToLower(protoForFamily(protocol, r.v6)),
 		"--dport", strconv.Itoa(int(sourcePort)),
 		"-d", localAddr.String(),
 		"-m", "addrtype", "--dst-type", "LOCAL",
@@ -990,10 +1010,22 @@ func applyPort(flag string, port *firewall.Port) []string {
 	return []string{flag, strconv.Itoa(int(port.Values[0]))}
 }
 
+// ipsetName returns the ipset name, suffixed with "-v6" for the v6 router
+// to avoid collisions since ipsets are global in the kernel.
+func (r *router) ipsetName(name string) string {
+	if r.v6 {
+		return name + "-v6"
+	}
+	return name
+}
+
 func (r *router) createIPSet(name string) error {
 	opts := ipset.CreateOptions{
 		Replace: true,
 	}
+	if r.v6 {
+		opts.Family = ipset.FamilyIPV6
+	}
 
 	if err := ipset.Create(name, ipset.TypeHashNet, opts); err != nil {
 		return fmt.Errorf("create ipset %s: %w", name, err)

client/firewall/iptables/rule.go

@@ -9,6 +9,7 @@ type Rule struct {
 	mangleSpecs []string
 	ip          string
 	chain       string
+	v6          bool
 }
 
 // GetRuleID returns the rule id

client/firewall/iptables/state_linux.go

@@ -4,6 +4,8 @@ import (
 	"fmt"
 	"sync"
 
+	log "github.com/sirupsen/logrus"
+
 	"github.com/netbirdio/netbird/client/iface"
 	"github.com/netbirdio/netbird/client/iface/wgaddr"
 )
@@ -37,6 +39,12 @@ type ShutdownState struct {
 
 	ACLEntries    aclEntries  `json:"acl_entries,omitempty"`
 	ACLIPsetStore *ipsetStore `json:"acl_ipset_store,omitempty"`
+
+	// IPv6 counterparts
+	RouteRules6        routeRules    `json:"route_rules_v6,omitempty"`
+	RouteIPsetCounter6 *ipsetCounter `json:"route_ipset_counter_v6,omitempty"`
+	ACLEntries6        aclEntries    `json:"acl_entries_v6,omitempty"`
+	ACLIPsetStore6     *ipsetStore   `json:"acl_ipset_store_v6,omitempty"`
 }
 
 func (s *ShutdownState) Name() string {
@@ -67,6 +75,28 @@ func (s *ShutdownState) Cleanup() error {
 		ipt.aclMgr.ipsetStore = s.ACLIPsetStore
 	}
 
+	// Clean up v6 state even if the current run has no IPv6.
+	// The previous run may have left ip6tables rules behind.
+	if !ipt.hasIPv6() {
+		if err := ipt.createIPv6Components(s.InterfaceState, mtu); err != nil {
+			log.Warnf("failed to create v6 components for cleanup: %v", err)
+		}
+	}
+	if ipt.hasIPv6() {
+		if s.RouteRules6 != nil {
+			ipt.router6.rules = s.RouteRules6
+		}
+		if s.RouteIPsetCounter6 != nil {
+			ipt.router6.ipsetCounter.LoadData(s.RouteIPsetCounter6)
+		}
+		if s.ACLEntries6 != nil {
+			ipt.aclMgr6.entries = s.ACLEntries6
+		}
+		if s.ACLIPsetStore6 != nil {
+			ipt.aclMgr6.ipsetStore = s.ACLIPsetStore6
+		}
+	}
+
 	if err := ipt.Close(nil); err != nil {
 		return fmt.Errorf("reset iptables manager: %w", err)
 	}

client/firewall/manager/firewall.go

@@ -168,6 +168,10 @@ type Manager interface {
 
 	// RemoveInboundDNAT removes inbound DNAT rule
 	RemoveInboundDNAT(localAddr netip.Addr, protocol Protocol, sourcePort, targetPort uint16) error
+
+	// SetupEBPFProxyNoTrack creates static notrack rules for eBPF proxy loopback traffic.
+	// This prevents conntrack from interfering with WireGuard proxy communication.
+	SetupEBPFProxyNoTrack(proxyPort, wgPort uint16) error
 }
 
 func GenKey(format string, pair RouterPair) string {

client/firewall/nftables/acl_linux.go

@@ -33,15 +33,12 @@ const (
 
 const flushError = "flush: %w"
 
-var (
-	anyIP = []byte{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
-)
-
 type AclManager struct {
 	rConn              *nftables.Conn
 	sConn              *nftables.Conn
 	wgIface            iFaceMapper
 	routingFwChainName string
+	af                 addrFamily
 
 	workTable       *nftables.Table
 	chainInputRules *nftables.Chain
@@ -67,6 +64,7 @@ func newAclManager(table *nftables.Table, wgIface iFaceMapper, routingFwChainNam
 		wgIface:            wgIface,
 		workTable:          table,
 		routingFwChainName: routingFwChainName,
+		af:                 familyForAddr(table.Family == nftables.TableFamilyIPv4),
 
 		ipsetStore: newIpsetStore(),
 		rules:      make(map[string]*Rule),
@@ -145,7 +143,7 @@ func (m *AclManager) DeletePeerRule(rule firewall.Rule) error {
 	}
 
 	if _, ok := ips[r.ip.String()]; ok {
-		err := m.sConn.SetDeleteElements(r.nftSet, []nftables.SetElement{{Key: r.ip.To4()}})
+		err := m.sConn.SetDeleteElements(r.nftSet, []nftables.SetElement{{Key: ipToBytes(r.ip, m.af)}})
 		if err != nil {
 			log.Errorf("delete elements for set %q: %v", r.nftSet.Name, err)
 		}
@@ -254,11 +252,11 @@ func (m *AclManager) addIOFiltering(
 		expressions = append(expressions, &expr.Payload{
 			DestRegister: 1,
 			Base:         expr.PayloadBaseNetworkHeader,
-			Offset:       uint32(9),
+			Offset:       m.af.protoOffset,
 			Len:          uint32(1),
 		})
 
-		protoData, err := protoToInt(proto)
+		protoData, err := m.af.protoNum(proto)
 		if err != nil {
 			return nil, fmt.Errorf("convert protocol to number: %v", err)
 		}
@@ -270,19 +268,16 @@ func (m *AclManager) addIOFiltering(
 		})
 	}
 
-	rawIP := ip.To4()
+	rawIP := ipToBytes(ip, m.af)
 	// check if rawIP contains zeroed IPv4 0.0.0.0 value
 	// in that case not add IP match expression into the rule definition
-	if !bytes.HasPrefix(anyIP, rawIP) {
-		// source address position
-		addrOffset := uint32(12)
-
+	if slices.ContainsFunc(rawIP, func(v byte) bool { return v != 0 }) {
 		expressions = append(expressions,
 			&expr.Payload{
 				DestRegister: 1,
 				Base:         expr.PayloadBaseNetworkHeader,
-				Offset:       addrOffset,
-				Len:          4,
+				Offset:       m.af.srcAddrOffset,
+				Len:          m.af.addrLen,
 			},
 		)
 		// add individual IP for match if no ipset defined
@@ -587,7 +582,7 @@ func (m *AclManager) addJumpRule(chain *nftables.Chain, to string, ifaceKey expr
 
 func (m *AclManager) addIpToSet(ipsetName string, ip net.IP) (*nftables.Set, error) {
 	ipset, err := m.rConn.GetSetByName(m.workTable, ipsetName)
-	rawIP := ip.To4()
+	rawIP := ipToBytes(ip, m.af)
 	if err != nil {
 		if ipset, err = m.createSet(m.workTable, ipsetName); err != nil {
 			return nil, fmt.Errorf("get set name: %v", err)
@@ -619,7 +614,7 @@ func (m *AclManager) createSet(table *nftables.Table, name string) (*nftables.Se
 		Name:    name,
 		Table:   table,
 		Dynamic: true,
-		KeyType: nftables.TypeIPAddr,
+		KeyType: m.af.setKeyType,
 	}
 
 	if err := m.rConn.AddSet(ipset, nil); err != nil {
@@ -707,15 +702,12 @@ func ifname(n string) []byte {
 	return b
 }
 
-func protoToInt(protocol firewall.Protocol) (uint8, error) {
-	switch protocol {
-	case firewall.ProtocolTCP:
-		return unix.IPPROTO_TCP, nil
-	case firewall.ProtocolUDP:
-		return unix.IPPROTO_UDP, nil
-	case firewall.ProtocolICMP:
-		return unix.IPPROTO_ICMP, nil
-	}
 
-	return 0, fmt.Errorf("unsupported protocol: %s", protocol)
+// ipToBytes converts net.IP to the correct byte length for the address family.
+func ipToBytes(ip net.IP, af addrFamily) []byte {
+	if af.addrLen == 4 {
+		return ip.To4()
+	}
+	return ip.To16()
 }
+

client/firewall/nftables/addr_family_linux.go

@@ -0,0 +1,81 @@
+package nftables
+
+import (
+	"fmt"
+	"net"
+
+	"github.com/google/nftables"
+	"golang.org/x/sys/unix"
+
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+)
+
+// addrFamily holds protocol-specific constants for nftables expression building.
+type addrFamily struct {
+	// protoOffset is the IP header offset for the protocol/next-header field (9 for v4, 6 for v6)
+	protoOffset uint32
+	// srcAddrOffset is the IP header offset for the source address (12 for v4, 8 for v6)
+	srcAddrOffset uint32
+	// dstAddrOffset is the IP header offset for the destination address (16 for v4, 24 for v6)
+	dstAddrOffset uint32
+	// addrLen is the byte length of addresses (4 for v4, 16 for v6)
+	addrLen uint32
+	// totalBits is the address size in bits (32 for v4, 128 for v6)
+	totalBits int
+	// setKeyType is the nftables set data type for addresses
+	setKeyType nftables.SetDatatype
+	// tableFamily is the nftables table family
+	tableFamily nftables.TableFamily
+	// icmpProto is the ICMP protocol number for this family (1 for v4, 58 for v6)
+	icmpProto uint8
+}
+
+var (
+	// afIPv4 defines IPv4 header layout and nftables types.
+	afIPv4 = addrFamily{
+		protoOffset:   9,
+		srcAddrOffset: 12,
+		dstAddrOffset: 16,
+		addrLen:       net.IPv4len,
+		totalBits:     8 * net.IPv4len,
+		setKeyType:    nftables.TypeIPAddr,
+		tableFamily:   nftables.TableFamilyIPv4,
+		icmpProto:     unix.IPPROTO_ICMP,
+	}
+	// afIPv6 defines IPv6 header layout and nftables types.
+	afIPv6 = addrFamily{
+		protoOffset:   6,
+		srcAddrOffset: 8,
+		dstAddrOffset: 24,
+		addrLen:       net.IPv6len,
+		totalBits:     8 * net.IPv6len,
+		setKeyType:    nftables.TypeIP6Addr,
+		tableFamily:   nftables.TableFamilyIPv6,
+		icmpProto:     unix.IPPROTO_ICMPV6,
+	}
+)
+
+// familyForAddr returns the address family for the given IP.
+func familyForAddr(is4 bool) addrFamily {
+	if is4 {
+		return afIPv4
+	}
+	return afIPv6
+}
+
+// protoNum converts a firewall protocol to the IP protocol number,
+// using the correct ICMP variant for the address family.
+func (af addrFamily) protoNum(protocol firewall.Protocol) (uint8, error) {
+	switch protocol {
+	case firewall.ProtocolTCP:
+		return unix.IPPROTO_TCP, nil
+	case firewall.ProtocolUDP:
+		return unix.IPPROTO_UDP, nil
+	case firewall.ProtocolICMP:
+		return af.icmpProto, nil
+	case firewall.ProtocolALL:
+		return 0, nil
+	default:
+		return 0, fmt.Errorf("unsupported protocol: %s", protocol)
+	}
+}

client/firewall/nftables/manager_linux.go

@@ -12,6 +12,7 @@ import (
 	"github.com/google/nftables/binaryutil"
 	"github.com/google/nftables/expr"
 	log "github.com/sirupsen/logrus"
+	"golang.org/x/sys/unix"
 
 	firewall "github.com/netbirdio/netbird/client/firewall/manager"
 	"github.com/netbirdio/netbird/client/iface/wgaddr"
@@ -50,6 +51,13 @@ type Manager struct {
 
 	router     *router
 	aclManager *AclManager
+
+	// IPv6 counterparts, nil when no v6 overlay
+	router6     *router
+	aclManager6 *AclManager
+
+	notrackOutputChain     *nftables.Chain
+	notrackPreroutingChain *nftables.Chain
 }
 
 // Create nftables firewall manager
@@ -59,7 +67,8 @@ func Create(wgIface iFaceMapper, mtu uint16) (*Manager, error) {
 		wgIface: wgIface,
 	}
 
-	workTable := &nftables.Table{Name: getTableName(), Family: nftables.TableFamilyIPv4}
+	tableName := getTableName()
+	workTable := &nftables.Table{Name: tableName, Family: nftables.TableFamilyIPv4}
 
 	var err error
 	m.router, err = newRouter(workTable, wgIface, mtu)
@@ -72,9 +81,54 @@ func Create(wgIface iFaceMapper, mtu uint16) (*Manager, error) {
 		return nil, fmt.Errorf("create acl manager: %w", err)
 	}
 
+	if wgIface.Address().HasIPv6() {
+		if err := m.createIPv6Components(tableName, wgIface, mtu); err != nil {
+			return nil, fmt.Errorf("create IPv6 firewall: %w", err)
+		}
+	}
+
 	return m, nil
 }
 
+func (m *Manager) createIPv6Components(tableName string, wgIface iFaceMapper, mtu uint16) error {
+	workTable6 := &nftables.Table{Name: tableName, Family: nftables.TableFamilyIPv6}
+
+	var err error
+	m.router6, err = newRouter(workTable6, wgIface, mtu)
+	if err != nil {
+		return fmt.Errorf("create v6 router: %w", err)
+	}
+
+	m.aclManager6, err = newAclManager(workTable6, wgIface, chainNameRoutingFw)
+	if err != nil {
+		return fmt.Errorf("create v6 acl manager: %w", err)
+	}
+
+	return nil
+}
+
+// hasIPv6 reports whether the manager has IPv6 components initialized.
+func (m *Manager) hasIPv6() bool {
+	return m.router6 != nil
+}
+
+func (m *Manager) initIPv6() error {
+	workTable6, err := m.createWorkTableFamily(nftables.TableFamilyIPv6)
+	if err != nil {
+		return fmt.Errorf("create v6 work table: %w", err)
+	}
+
+	if err := m.router6.init(workTable6); err != nil {
+		return fmt.Errorf("v6 router init: %w", err)
+	}
+
+	if err := m.aclManager6.init(workTable6); err != nil {
+		return fmt.Errorf("v6 acl manager init: %w", err)
+	}
+
+	return nil
+}
+
 // Init nftables firewall manager
 func (m *Manager) Init(stateManager *statemanager.Manager) error {
 	workTable, err := m.createWorkTable()
@@ -91,6 +145,17 @@ func (m *Manager) Init(stateManager *statemanager.Manager) error {
 		return fmt.Errorf("acl manager init: %w", err)
 	}
 
+	if m.hasIPv6() {
+		if err := m.initIPv6(); err != nil {
+			// Peer has a v6 address: v6 firewall MUST work or we risk fail-open.
+			return fmt.Errorf("init IPv6 firewall (required because peer has IPv6 address): %w", err)
+		}
+	}
+
+	if err := m.initNoTrackChains(workTable); err != nil {
+		log.Warnf("raw priority chains not available, notrack rules will be disabled: %v", err)
+	}
+
 	stateManager.RegisterState(&ShutdownState{})
 
 	// We only need to record minimal interface state for potential recreation.
@@ -134,12 +199,14 @@ func (m *Manager) AddPeerFiltering(
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	rawIP := ip.To4()
-	if rawIP == nil {
-		return nil, fmt.Errorf("unsupported IP version: %s", ip.String())
+	if ip.To4() != nil {
+		return m.aclManager.AddPeerFiltering(id, ip, proto, sPort, dPort, action, ipsetName)
 	}
 
-	return m.aclManager.AddPeerFiltering(id, ip, proto, sPort, dPort, action, ipsetName)
+	if !m.hasIPv6() {
+		return nil, fmt.Errorf("IPv6 not initialized, cannot add rule for %s", ip)
+	}
+	return m.aclManager6.AddPeerFiltering(id, ip, proto, sPort, dPort, action, ipsetName)
 }
 
 func (m *Manager) AddRouteFiltering(
@@ -153,8 +220,11 @@ func (m *Manager) AddRouteFiltering(
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	if destination.IsPrefix() && !destination.Prefix.Addr().Is4() {
-		return nil, fmt.Errorf("unsupported IP version: %s", destination.Prefix.Addr().String())
+	if isIPv6RouteRule(sources, destination) {
+		if !m.hasIPv6() {
+			return nil, fmt.Errorf("IPv6 not initialized, cannot add route rule")
+		}
+		return m.router6.AddRouteFiltering(id, sources, destination, proto, sPort, dPort, action)
 	}
 
 	return m.router.AddRouteFiltering(id, sources, destination, proto, sPort, dPort, action)
@@ -165,14 +235,38 @@ func (m *Manager) DeletePeerRule(rule firewall.Rule) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && isIPv6Rule(rule) {
+		return m.aclManager6.DeletePeerRule(rule)
+	}
 	return m.aclManager.DeletePeerRule(rule)
 }
 
-// DeleteRouteRule deletes a routing rule
+func isIPv6Rule(rule firewall.Rule) bool {
+	r, ok := rule.(*Rule)
+	return ok && r.nftRule != nil && r.nftRule.Table != nil && r.nftRule.Table.Family == nftables.TableFamilyIPv6
+}
+
+// isIPv6RouteRule determines whether a route rule belongs to the v6 table.
+// For static routes, the destination prefix determines the family. For dynamic
+// routes (DomainSet), the sources determine the family since management
+// duplicates dynamic rules per family.
+func isIPv6RouteRule(sources []netip.Prefix, destination firewall.Network) bool {
+	if destination.IsPrefix() {
+		return destination.Prefix.Addr().Is6()
+	}
+	return len(sources) > 0 && sources[0].Addr().Is6()
+}
+
+// DeleteRouteRule deletes a routing rule.
+// Route rules are keyed by content hash, so the rule exists in exactly one
+// router. We check v4 first; if the key isn't there, try v6.
 func (m *Manager) DeleteRouteRule(rule firewall.Rule) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && !m.router.hasRule(rule.ID()) {
+		return m.router6.DeleteRouteRule(rule)
+	}
 	return m.router.DeleteRouteRule(rule)
 }
 
@@ -188,17 +282,63 @@ func (m *Manager) AddNatRule(pair firewall.RouterPair) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.AddNatRule(pair)
+	if pair.Destination.IsPrefix() && pair.Destination.Prefix.Addr().Is6() {
+		if !m.hasIPv6() {
+			return fmt.Errorf("IPv6 not initialized, cannot add NAT rule")
+		}
+		return m.router6.AddNatRule(pair)
+	}
+
+	if err := m.router.AddNatRule(pair); err != nil {
+		return err
+	}
+
+	// Dynamic routes (DomainSet) need NAT in both tables since resolved IPs
+	// can be either v4 or v6.
+	if m.hasIPv6() && pair.Destination.IsSet() {
+		v6Pair := pair
+		v6Pair.Source = firewall.Network{Prefix: netip.PrefixFrom(netip.IPv6Unspecified(), 0)}
+		if err := m.router6.AddNatRule(v6Pair); err != nil {
+			return fmt.Errorf("add v6 NAT rule: %w", err)
+		}
+	}
+
+	return nil
 }
 
 func (m *Manager) RemoveNatRule(pair firewall.RouterPair) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.RemoveNatRule(pair)
+	if pair.Destination.IsPrefix() && pair.Destination.Prefix.Addr().Is6() {
+		if !m.hasIPv6() {
+			return nil
+		}
+		return m.router6.RemoveNatRule(pair)
+	}
+
+	if err := m.router.RemoveNatRule(pair); err != nil {
+		return err
+	}
+
+	if m.hasIPv6() && pair.Destination.IsSet() {
+		v6Pair := pair
+		v6Pair.Source = firewall.Network{Prefix: netip.PrefixFrom(netip.IPv6Unspecified(), 0)}
+		if err := m.router6.RemoveNatRule(v6Pair); err != nil {
+			return fmt.Errorf("remove v6 NAT rule: %w", err)
+		}
+	}
+
+	return nil
 }
 
-// AllowNetbird allows netbird interface traffic
+// AllowNetbird allows netbird interface traffic.
+// TODO: In USP mode this only adds ACCEPT to the netbird table's own chains,
+// which doesn't override DROP rules in external tables (e.g. firewalld).
+// Should add passthrough rules to external chains (like the native mode router's
+// addExternalChainsRules does) for both the netbird table family and inet tables.
+// The netbird table itself is fine (routing chains already exist there), but
+// non-netbird tables with INPUT/FORWARD hooks can still DROP our WG traffic.
 func (m *Manager) AllowNetbird() error {
 	if !m.wgIface.IsUserspaceBind() {
 		return nil
@@ -210,6 +350,11 @@ func (m *Manager) AllowNetbird() error {
 	if err := m.aclManager.createDefaultAllowRules(); err != nil {
 		return fmt.Errorf("create default allow rules: %w", err)
 	}
+	if m.hasIPv6() {
+		if err := m.aclManager6.createDefaultAllowRules(); err != nil {
+			return fmt.Errorf("create v6 default allow rules: %w", err)
+		}
+	}
 	if err := m.rConn.Flush(); err != nil {
 		return fmt.Errorf("flush allow input netbird rules: %w", err)
 	}
@@ -219,7 +364,13 @@ func (m *Manager) AllowNetbird() error {
 
 // SetLegacyManagement sets the route manager to use legacy management
 func (m *Manager) SetLegacyManagement(isLegacy bool) error {
-	return firewall.SetLegacyManagement(m.router, isLegacy)
+	if err := firewall.SetLegacyManagement(m.router, isLegacy); err != nil {
+		return err
+	}
+	if m.hasIPv6() {
+		return firewall.SetLegacyManagement(m.router6, isLegacy)
+	}
+	return nil
 }
 
 // Close closes the firewall manager
@@ -231,6 +382,12 @@ func (m *Manager) Close(stateManager *statemanager.Manager) error {
 		return fmt.Errorf("reset router: %v", err)
 	}
 
+	if m.hasIPv6() {
+		if err := m.router6.Reset(); err != nil {
+			return fmt.Errorf("reset v6 router: %v", err)
+		}
+	}
+
 	if err := m.cleanupNetbirdTables(); err != nil {
 		return fmt.Errorf("cleanup netbird tables: %v", err)
 	}
@@ -288,7 +445,21 @@ func (m *Manager) Flush() error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.aclManager.Flush()
+	if err := m.aclManager.Flush(); err != nil {
+		return err
+	}
+
+	if m.hasIPv6() {
+		if err := m.aclManager6.Flush(); err != nil {
+			return fmt.Errorf("flush v6 acl: %w", err)
+		}
+	}
+
+	if err := m.refreshNoTrackChains(); err != nil {
+		log.Errorf("failed to refresh notrack chains: %v", err)
+	}
+
+	return nil
 }
 
 // AddDNATRule adds a DNAT rule
@@ -296,6 +467,9 @@ func (m *Manager) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error)
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && rule.TranslatedAddress.Is6() {
+		return m.router6.AddDNATRule(rule)
+	}
 	return m.router.AddDNATRule(rule)
 }
 
@@ -304,6 +478,9 @@ func (m *Manager) DeleteDNATRule(rule firewall.Rule) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && !m.router.hasDNATRule(rule.ID()) {
+		return m.router6.DeleteDNATRule(rule)
+	}
 	return m.router.DeleteDNATRule(rule)
 }
 
@@ -312,7 +489,26 @@ func (m *Manager) UpdateSet(set firewall.Set, prefixes []netip.Prefix) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.UpdateSet(set, prefixes)
+	var v4Prefixes, v6Prefixes []netip.Prefix
+	for _, p := range prefixes {
+		if p.Addr().Is6() {
+			v6Prefixes = append(v6Prefixes, p)
+		} else {
+			v4Prefixes = append(v4Prefixes, p)
+		}
+	}
+
+	if err := m.router.UpdateSet(set, v4Prefixes); err != nil {
+		return err
+	}
+
+	if m.hasIPv6() && len(v6Prefixes) > 0 {
+		if err := m.router6.UpdateSet(set, v6Prefixes); err != nil {
+			return fmt.Errorf("update v6 set: %w", err)
+		}
+	}
+
+	return nil
 }
 
 // AddInboundDNAT adds an inbound DNAT rule redirecting traffic from NetBird peers to local services.
@@ -320,6 +516,9 @@ func (m *Manager) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protoco
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && localAddr.Is6() {
+		return m.router6.AddInboundDNAT(localAddr, protocol, sourcePort, targetPort)
+	}
 	return m.router.AddInboundDNAT(localAddr, protocol, sourcePort, targetPort)
 }
 
@@ -328,11 +527,188 @@ func (m *Manager) RemoveInboundDNAT(localAddr netip.Addr, protocol firewall.Prot
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
+	if m.hasIPv6() && localAddr.Is6() {
+		return m.router6.RemoveInboundDNAT(localAddr, protocol, sourcePort, targetPort)
+	}
 	return m.router.RemoveInboundDNAT(localAddr, protocol, sourcePort, targetPort)
 }
 
+const (
+	chainNameRawOutput     = "netbird-raw-out"
+	chainNameRawPrerouting = "netbird-raw-pre"
+)
+
+// SetupEBPFProxyNoTrack creates notrack rules for eBPF proxy loopback traffic.
+// This prevents conntrack from tracking WireGuard proxy traffic on loopback, which
+// can interfere with MASQUERADE rules (e.g., from container runtimes like Podman/netavark).
+//
+// Traffic flows that need NOTRACK:
+//
+//  1. Egress: WireGuard -> fake endpoint (before eBPF rewrite)
+//     src=127.0.0.1:wgPort -> dst=127.0.0.1:fakePort
+//     Matched by: sport=wgPort
+//
+//  2. Egress: Proxy -> WireGuard (via raw socket)
+//     src=127.0.0.1:fakePort -> dst=127.0.0.1:wgPort
+//     Matched by: dport=wgPort
+//
+//  3. Ingress: Packets to WireGuard
+//     dst=127.0.0.1:wgPort
+//     Matched by: dport=wgPort
+//
+//  4. Ingress: Packets to proxy (after eBPF rewrite)
+//     dst=127.0.0.1:proxyPort
+//     Matched by: dport=proxyPort
+//
+// Rules are cleaned up when the firewall manager is closed.
+func (m *Manager) SetupEBPFProxyNoTrack(proxyPort, wgPort uint16) error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	if m.notrackOutputChain == nil || m.notrackPreroutingChain == nil {
+		return fmt.Errorf("notrack chains not initialized")
+	}
+
+	proxyPortBytes := binaryutil.BigEndian.PutUint16(proxyPort)
+	wgPortBytes := binaryutil.BigEndian.PutUint16(wgPort)
+	loopback := []byte{127, 0, 0, 1}
+
+	// Egress rules: match outgoing loopback UDP packets
+	m.rConn.AddRule(&nftables.Rule{
+		Table: m.notrackOutputChain.Table,
+		Chain: m.notrackOutputChain,
+		Exprs: []expr.Any{
+			&expr.Meta{Key: expr.MetaKeyOIFNAME, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: ifname("lo")},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseNetworkHeader, Offset: 12, Len: 4}, // saddr
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: loopback},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseNetworkHeader, Offset: 16, Len: 4}, // daddr
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: loopback},
+			&expr.Meta{Key: expr.MetaKeyL4PROTO, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: []byte{unix.IPPROTO_UDP}},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseTransportHeader, Offset: 0, Len: 2},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: wgPortBytes}, // sport=wgPort
+			&expr.Counter{},
+			&expr.Notrack{},
+		},
+	})
+	m.rConn.AddRule(&nftables.Rule{
+		Table: m.notrackOutputChain.Table,
+		Chain: m.notrackOutputChain,
+		Exprs: []expr.Any{
+			&expr.Meta{Key: expr.MetaKeyOIFNAME, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: ifname("lo")},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseNetworkHeader, Offset: 12, Len: 4}, // saddr
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: loopback},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseNetworkHeader, Offset: 16, Len: 4}, // daddr
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: loopback},
+			&expr.Meta{Key: expr.MetaKeyL4PROTO, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: []byte{unix.IPPROTO_UDP}},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseTransportHeader, Offset: 2, Len: 2},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: wgPortBytes}, // dport=wgPort
+			&expr.Counter{},
+			&expr.Notrack{},
+		},
+	})
+
+	// Ingress rules: match incoming loopback UDP packets
+	m.rConn.AddRule(&nftables.Rule{
+		Table: m.notrackPreroutingChain.Table,
+		Chain: m.notrackPreroutingChain,
+		Exprs: []expr.Any{
+			&expr.Meta{Key: expr.MetaKeyIIFNAME, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: ifname("lo")},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseNetworkHeader, Offset: 12, Len: 4}, // saddr
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: loopback},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseNetworkHeader, Offset: 16, Len: 4}, // daddr
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: loopback},
+			&expr.Meta{Key: expr.MetaKeyL4PROTO, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: []byte{unix.IPPROTO_UDP}},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseTransportHeader, Offset: 2, Len: 2},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: wgPortBytes}, // dport=wgPort
+			&expr.Counter{},
+			&expr.Notrack{},
+		},
+	})
+	m.rConn.AddRule(&nftables.Rule{
+		Table: m.notrackPreroutingChain.Table,
+		Chain: m.notrackPreroutingChain,
+		Exprs: []expr.Any{
+			&expr.Meta{Key: expr.MetaKeyIIFNAME, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: ifname("lo")},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseNetworkHeader, Offset: 12, Len: 4}, // saddr
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: loopback},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseNetworkHeader, Offset: 16, Len: 4}, // daddr
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: loopback},
+			&expr.Meta{Key: expr.MetaKeyL4PROTO, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: []byte{unix.IPPROTO_UDP}},
+			&expr.Payload{DestRegister: 1, Base: expr.PayloadBaseTransportHeader, Offset: 2, Len: 2},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: proxyPortBytes}, // dport=proxyPort
+			&expr.Counter{},
+			&expr.Notrack{},
+		},
+	})
+
+	if err := m.rConn.Flush(); err != nil {
+		return fmt.Errorf("flush notrack rules: %w", err)
+	}
+
+	log.Debugf("set up ebpf proxy notrack rules for ports %d,%d", proxyPort, wgPort)
+	return nil
+}
+
+func (m *Manager) initNoTrackChains(table *nftables.Table) error {
+	m.notrackOutputChain = m.rConn.AddChain(&nftables.Chain{
+		Name:     chainNameRawOutput,
+		Table:    table,
+		Type:     nftables.ChainTypeFilter,
+		Hooknum:  nftables.ChainHookOutput,
+		Priority: nftables.ChainPriorityRaw,
+	})
+
+	m.notrackPreroutingChain = m.rConn.AddChain(&nftables.Chain{
+		Name:     chainNameRawPrerouting,
+		Table:    table,
+		Type:     nftables.ChainTypeFilter,
+		Hooknum:  nftables.ChainHookPrerouting,
+		Priority: nftables.ChainPriorityRaw,
+	})
+
+	if err := m.rConn.Flush(); err != nil {
+		return fmt.Errorf("flush chain creation: %w", err)
+	}
+
+	return nil
+}
+
+func (m *Manager) refreshNoTrackChains() error {
+	chains, err := m.rConn.ListChainsOfTableFamily(nftables.TableFamilyIPv4)
+	if err != nil {
+		return fmt.Errorf("list chains: %w", err)
+	}
+
+	tableName := getTableName()
+	for _, c := range chains {
+		if c.Table.Name != tableName {
+			continue
+		}
+		switch c.Name {
+		case chainNameRawOutput:
+			m.notrackOutputChain = c
+		case chainNameRawPrerouting:
+			m.notrackPreroutingChain = c
+		}
+	}
+
+	return nil
+}
+
 func (m *Manager) createWorkTable() (*nftables.Table, error) {
-	tables, err := m.rConn.ListTablesOfFamily(nftables.TableFamilyIPv4)
+	return m.createWorkTableFamily(nftables.TableFamilyIPv4)
+}
+
+func (m *Manager) createWorkTableFamily(family nftables.TableFamily) (*nftables.Table, error) {
+	tables, err := m.rConn.ListTablesOfFamily(family)
 	if err != nil {
 		return nil, fmt.Errorf("list of tables: %w", err)
 	}
@@ -344,7 +720,7 @@ func (m *Manager) createWorkTable() (*nftables.Table, error) {
 		}
 	}
 
-	table := m.rConn.AddTable(&nftables.Table{Name: getTableName(), Family: nftables.TableFamilyIPv4})
+	table := m.rConn.AddTable(&nftables.Table{Name: tableName, Family: family})
 	err = m.rConn.Flush()
 	return table, err
 }

client/firewall/nftables/manager_linux_test.go

@@ -198,7 +198,7 @@ func TestNftablesManagerRuleOrder(t *testing.T) {
 	t.Logf("Found %d rules in nftables chain", len(rules))
 
 	// Find the accept and deny rules and verify deny comes before accept
-	var acceptRuleIndex, denyRuleIndex int = -1, -1
+	var acceptRuleIndex, denyRuleIndex = -1, -1
 	for i, rule := range rules {
 		hasAcceptHTTPSet := false
 		hasDenyHTTPSet := false
@@ -208,11 +208,13 @@ func TestNftablesManagerRuleOrder(t *testing.T) {
 		for _, e := range rule.Exprs {
 			// Check for set lookup
 			if lookup, ok := e.(*expr.Lookup); ok {
-				if lookup.SetName == "accept-http" {
+				switch lookup.SetName {
+				case "accept-http":
 					hasAcceptHTTPSet = true
-				} else if lookup.SetName == "deny-http" {
+				case "deny-http":
 					hasDenyHTTPSet = true
 				}
+
 			}
 			// Check for port 80
 			if cmp, ok := e.(*expr.Cmp); ok {
@@ -222,9 +224,10 @@ func TestNftablesManagerRuleOrder(t *testing.T) {
 			}
 			// Check for verdict
 			if verdict, ok := e.(*expr.Verdict); ok {
-				if verdict.Kind == expr.VerdictAccept {
+				switch verdict.Kind {
+				case expr.VerdictAccept:
 					action = "ACCEPT"
-				} else if verdict.Kind == expr.VerdictDrop {
+				case expr.VerdictDrop:
 					action = "DROP"
 				}
 			}
@@ -382,10 +385,132 @@ func TestNftablesManagerCompatibilityWithIptables(t *testing.T) {
 	err = manager.AddNatRule(pair)
 	require.NoError(t, err, "failed to add NAT rule")
 
+	dnatRule, err := manager.AddDNATRule(fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{8080}},
+		TranslatedAddress: netip.MustParseAddr("100.96.0.2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	})
+	require.NoError(t, err, "failed to add DNAT rule")
+
+	t.Cleanup(func() {
+		require.NoError(t, manager.DeleteDNATRule(dnatRule), "failed to delete DNAT rule")
+	})
+
 	stdout, stderr = runIptablesSave(t)
 	verifyIptablesOutput(t, stdout, stderr)
 }
 
+func TestNftablesManagerIPv6CompatibilityWithIp6tables(t *testing.T) {
+	if check() != NFTABLES {
+		t.Skip("nftables not supported on this system")
+	}
+
+	if _, err := exec.LookPath("ip6tables-save"); err != nil {
+		t.Skipf("ip6tables-save not available on this system: %v", err)
+	}
+
+	// Seed ip6 tables in the nft backend. Docker may not create them.
+	seedIp6tables(t)
+
+	ifaceMockV6 := &iFaceMock{
+		NameFunc: func() string { return "wt-test" },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("100.96.0.1"),
+				Network: netip.MustParsePrefix("100.96.0.0/16"),
+				IPv6:    netip.MustParseAddr("fd00::1"),
+				IPv6Net: netip.MustParsePrefix("fd00::/64"),
+			}
+		},
+	}
+
+	manager, err := Create(ifaceMockV6, iface.DefaultMTU)
+	require.NoError(t, err, "create manager")
+	require.NoError(t, manager.Init(nil))
+
+	t.Cleanup(func() {
+		require.NoError(t, manager.Close(nil), "close manager")
+
+		stdout, stderr := runIp6tablesSave(t)
+		verifyIp6tablesOutput(t, stdout, stderr)
+	})
+
+	ip := netip.MustParseAddr("fd00::2")
+	_, err = manager.AddPeerFiltering(nil, ip.AsSlice(), fw.ProtocolTCP, nil, &fw.Port{Values: []uint16{80}}, fw.ActionAccept, "")
+	require.NoError(t, err, "add v6 peer filtering rule")
+
+	_, err = manager.AddRouteFiltering(
+		nil,
+		[]netip.Prefix{netip.MustParsePrefix("fd00:1::/64")},
+		fw.Network{Prefix: netip.MustParsePrefix("2001:db8::/48")},
+		fw.ProtocolTCP,
+		nil,
+		&fw.Port{Values: []uint16{443}},
+		fw.ActionAccept,
+	)
+	require.NoError(t, err, "add v6 route filtering rule")
+
+	err = manager.AddNatRule(fw.RouterPair{
+		Source:      fw.Network{Prefix: netip.MustParsePrefix("fd00::/64")},
+		Destination: fw.Network{Prefix: netip.MustParsePrefix("2001:db8::/48")},
+		Masquerade:  true,
+	})
+	require.NoError(t, err, "add v6 NAT rule")
+
+	dnatRule, err := manager.AddDNATRule(fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{8080}},
+		TranslatedAddress: netip.MustParseAddr("fd00::2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	})
+	require.NoError(t, err, "add v6 DNAT rule")
+
+	t.Cleanup(func() {
+		require.NoError(t, manager.DeleteDNATRule(dnatRule), "delete v6 DNAT rule")
+	})
+
+	stdout, stderr := runIptablesSave(t)
+	verifyIptablesOutput(t, stdout, stderr)
+
+	stdout, stderr = runIp6tablesSave(t)
+	verifyIp6tablesOutput(t, stdout, stderr)
+}
+
+func seedIp6tables(t *testing.T) {
+	t.Helper()
+	for _, tc := range []struct{ table, chain string }{
+		{"filter", "FORWARD"},
+		{"nat", "POSTROUTING"},
+		{"mangle", "FORWARD"},
+	} {
+		add := exec.Command("ip6tables", "-t", tc.table, "-A", tc.chain, "-j", "ACCEPT")
+		require.NoError(t, add.Run(), "seed ip6tables -t %s", tc.table)
+		del := exec.Command("ip6tables", "-t", tc.table, "-D", tc.chain, "-j", "ACCEPT")
+		require.NoError(t, del.Run(), "unseed ip6tables -t %s", tc.table)
+	}
+}
+
+func runIp6tablesSave(t *testing.T) (string, string) {
+	t.Helper()
+	var stdout, stderr bytes.Buffer
+	cmd := exec.Command("ip6tables-save")
+	cmd.Stdout = &stdout
+	cmd.Stderr = &stderr
+	require.NoError(t, cmd.Run(), "ip6tables-save failed")
+	return stdout.String(), stderr.String()
+}
+
+func verifyIp6tablesOutput(t *testing.T, stdout, stderr string) {
+	t.Helper()
+	require.NotContains(t, stdout, "Table `nat' is incompatible",
+		"ip6tables-save: nat table incompatible. Full output: %s", stdout)
+	require.NotContains(t, stdout, "Table `mangle' is incompatible",
+		"ip6tables-save: mangle table incompatible. Full output: %s", stdout)
+	require.NotContains(t, stdout, "Table `filter' is incompatible",
+		"ip6tables-save: filter table incompatible. Full output: %s", stdout)
+}
+
 func TestNftablesManagerCompatibilityWithIptablesFor6kPrefixes(t *testing.T) {
 	if check() != NFTABLES {
 		t.Skip("nftables not supported on this system")

client/firewall/nftables/router_linux.go

@@ -46,8 +46,10 @@ const (
 	dnatSuffix = "_dnat"
 	snatSuffix = "_snat"
 
-	// ipTCPHeaderMinSize represents minimum IP (20) + TCP (20) header size for MSS calculation
-	ipTCPHeaderMinSize = 40
+	// ipv4TCPHeaderSize is the minimum IPv4 (20) + TCP (20) header size for MSS calculation.
+	ipv4TCPHeaderSize = 40
+	// ipv6TCPHeaderSize is the minimum IPv6 (40) + TCP (20) header size for MSS calculation.
+	ipv6TCPHeaderSize = 60
 
 	// maxPrefixesSet 1638 prefixes start to fail, taking some margin
 	maxPrefixesSet       = 1500
@@ -72,6 +74,7 @@ type router struct {
 	rules        map[string]*nftables.Rule
 	ipsetCounter *refcounter.Counter[string, setInput, *nftables.Set]
 
+	af               addrFamily
 	wgIface          iFaceMapper
 	ipFwdState       *ipfwdstate.IPForwardingState
 	legacyManagement bool
@@ -84,6 +87,7 @@ func newRouter(workTable *nftables.Table, wgIface iFaceMapper, mtu uint16) (*rou
 		workTable:  workTable,
 		chains:     make(map[string]*nftables.Chain),
 		rules:      make(map[string]*nftables.Rule),
+		af:         familyForAddr(workTable.Family == nftables.TableFamilyIPv4),
 		wgIface:    wgIface,
 		ipFwdState: ipfwdstate.NewIPForwardingState(),
 		mtu:        mtu,
@@ -142,7 +146,7 @@ func (r *router) Reset() error {
 func (r *router) removeNatPreroutingRules() error {
 	table := &nftables.Table{
 		Name:   tableNat,
-		Family: nftables.TableFamilyIPv4,
+		Family: r.af.tableFamily,
 	}
 	chain := &nftables.Chain{
 		Name:     chainNameNatPrerouting,
@@ -175,7 +179,7 @@ func (r *router) removeNatPreroutingRules() error {
 }
 
 func (r *router) loadFilterTable() (*nftables.Table, error) {
-	tables, err := r.conn.ListTablesOfFamily(nftables.TableFamilyIPv4)
+	tables, err := r.conn.ListTablesOfFamily(r.af.tableFamily)
 	if err != nil {
 		return nil, fmt.Errorf("list tables: %w", err)
 	}
@@ -407,7 +411,7 @@ func (r *router) AddRouteFiltering(
 
 	// Handle protocol
 	if proto != firewall.ProtocolALL {
-		protoNum, err := protoToInt(proto)
+		protoNum, err := r.af.protoNum(proto)
 		if err != nil {
 			return nil, fmt.Errorf("convert protocol to number: %w", err)
 		}
@@ -467,7 +471,24 @@ func (r *router) getIpSet(set firewall.Set, prefixes []netip.Prefix, isSource bo
 		return nil, fmt.Errorf("create or get ipset: %w", err)
 	}
 
-	return getIpSetExprs(ref, isSource)
+	return r.getIpSetExprs(ref, isSource)
+}
+
+func (r *router) iptablesProto() iptables.Protocol {
+	if r.af.tableFamily == nftables.TableFamilyIPv6 {
+		return iptables.ProtocolIPv6
+	}
+	return iptables.ProtocolIPv4
+}
+
+func (r *router) hasRule(id string) bool {
+	_, ok := r.rules[id]
+	return ok
+}
+
+func (r *router) hasDNATRule(id string) bool {
+	_, ok := r.rules[id+dnatSuffix]
+	return ok
 }
 
 func (r *router) DeleteRouteRule(rule firewall.Rule) error {
@@ -483,7 +504,12 @@ func (r *router) DeleteRouteRule(rule firewall.Rule) error {
 	}
 
 	if nftRule.Handle == 0 {
-		return fmt.Errorf("route rule %s has no handle", ruleKey)
+		log.Warnf("route rule %s has no handle, removing stale entry", ruleKey)
+		if err := r.decrementSetCounter(nftRule); err != nil {
+			log.Warnf("decrement set counter for stale rule %s: %v", ruleKey, err)
+		}
+		delete(r.rules, ruleKey)
+		return nil
 	}
 
 	if err := r.deleteNftRule(nftRule, ruleKey); err != nil {
@@ -511,10 +537,10 @@ func (r *router) createIpSet(setName string, input setInput) (*nftables.Set, err
 		Table:   r.workTable,
 		// required for prefixes
 		Interval: true,
-		KeyType:  nftables.TypeIPAddr,
+		KeyType:  r.af.setKeyType,
 	}
 
-	elements := convertPrefixesToSet(prefixes)
+	elements := r.convertPrefixesToSet(prefixes)
 	nElements := len(elements)
 
 	maxElements := maxPrefixesSet * 2
@@ -547,23 +573,17 @@ func (r *router) createIpSet(setName string, input setInput) (*nftables.Set, err
 	return nfset, nil
 }
 
-func convertPrefixesToSet(prefixes []netip.Prefix) []nftables.SetElement {
+func (r *router) convertPrefixesToSet(prefixes []netip.Prefix) []nftables.SetElement {
 	var elements []nftables.SetElement
 	for _, prefix := range prefixes {
-		// TODO: Implement IPv6 support
-		if prefix.Addr().Is6() {
-			log.Tracef("skipping IPv6 prefix %s: IPv6 support not yet implemented", prefix)
-			continue
-		}
-
 		// nftables needs half-open intervals [firstIP, lastIP) for prefixes
 		// e.g. 10.0.0.0/24 becomes [10.0.0.0, 10.0.1.0), 10.1.1.1/32 becomes [10.1.1.1, 10.1.1.2) etc
 		firstIP := prefix.Addr()
 		lastIP := calculateLastIP(prefix).Next()
 
 		elements = append(elements,
-			// the nft tool also adds a line like this, see https://github.com/google/nftables/issues/247
-			// nftables.SetElement{Key: []byte{0, 0, 0, 0}, IntervalEnd: true},
+			// the nft tool also adds a zero-address IntervalEnd element, see https://github.com/google/nftables/issues/247
+			// nftables.SetElement{Key: make([]byte, r.af.addrLen), IntervalEnd: true},
 			nftables.SetElement{Key: firstIP.AsSlice()},
 			nftables.SetElement{Key: lastIP.AsSlice(), IntervalEnd: true},
 		)
@@ -573,10 +593,20 @@ func convertPrefixesToSet(prefixes []netip.Prefix) []nftables.SetElement {
 
 // calculateLastIP determines the last IP in a given prefix.
 func calculateLastIP(prefix netip.Prefix) netip.Addr {
-	hostMask := ^uint32(0) >> prefix.Masked().Bits()
-	lastIP := uint32FromNetipAddr(prefix.Addr()) | hostMask
+	masked := prefix.Masked()
+	if masked.Addr().Is4() {
+		hostMask := ^uint32(0) >> masked.Bits()
+		lastIP := uint32FromNetipAddr(masked.Addr()) | hostMask
+		return netip.AddrFrom4(uint32ToBytes(lastIP))
+	}
 
-	return netip.AddrFrom4(uint32ToBytes(lastIP))
+	// IPv6: set host bits to all 1s
+	b := masked.Addr().As16()
+	bits := masked.Bits()
+	for i := bits; i < 128; i++ {
+		b[i/8] |= 1 << (7 - i%8)
+	}
+	return netip.AddrFrom16(b)
 }
 
 // Utility function to convert netip.Addr to uint32.
@@ -660,13 +690,32 @@ func (r *router) AddNatRule(pair firewall.RouterPair) error {
 	}
 
 	if err := r.conn.Flush(); err != nil {
-		// TODO: rollback ipset counter
-		return fmt.Errorf("insert rules for %s: %v", pair.Destination, err)
+		r.rollbackRules(pair)
+		return fmt.Errorf("insert rules for %s: %w", pair.Destination, err)
 	}
 
 	return nil
 }
 
+// rollbackRules cleans up unflushed rules and their set counters after a flush failure.
+func (r *router) rollbackRules(pair firewall.RouterPair) {
+	keys := []string{
+		firewall.GenKey(firewall.ForwardingFormat, pair),
+		firewall.GenKey(firewall.PreroutingFormat, pair),
+		firewall.GenKey(firewall.PreroutingFormat, firewall.GetInversePair(pair)),
+	}
+	for _, key := range keys {
+		rule, ok := r.rules[key]
+		if !ok {
+			continue
+		}
+		if err := r.decrementSetCounter(rule); err != nil {
+			log.Warnf("rollback set counter for %s: %v", key, err)
+		}
+		delete(r.rules, key)
+	}
+}
+
 // addNatRule inserts a nftables rule to the conn client flush queue
 func (r *router) addNatRule(pair firewall.RouterPair) error {
 	sourceExp, err := r.applyNetwork(pair.Source, nil, true)
@@ -809,9 +858,12 @@ func (r *router) addPostroutingRules() {
 }
 
 // addMSSClampingRules adds MSS clamping rules to prevent fragmentation for forwarded traffic.
-// TODO: Add IPv6 support
 func (r *router) addMSSClampingRules() error {
-	mss := r.mtu - ipTCPHeaderMinSize
+	overhead := uint16(ipv4TCPHeaderSize)
+	if r.af.tableFamily == nftables.TableFamilyIPv6 {
+		overhead = ipv6TCPHeaderSize
+	}
+	mss := r.mtu - overhead
 
 	exprsOut := []expr.Any{
 		&expr.Meta{
@@ -928,18 +980,30 @@ func (r *router) addLegacyRouteRule(pair firewall.RouterPair) error {
 func (r *router) removeLegacyRouteRule(pair firewall.RouterPair) error {
 	ruleKey := firewall.GenKey(firewall.ForwardingFormat, pair)
 
-	if rule, exists := r.rules[ruleKey]; exists {
-		if err := r.conn.DelRule(rule); err != nil {
-			return fmt.Errorf("remove legacy forwarding rule %s -> %s: %v", pair.Source, pair.Destination, err)
-		}
-
-		log.Debugf("removed legacy forwarding rule %s -> %s", pair.Source, pair.Destination)
-
-		delete(r.rules, ruleKey)
+	rule, exists := r.rules[ruleKey]
+	if !exists {
+		return nil
+	}
 
+	if rule.Handle == 0 {
+		log.Warnf("legacy forwarding rule %s has no handle, removing stale entry", ruleKey)
 		if err := r.decrementSetCounter(rule); err != nil {
-			return fmt.Errorf("decrement set counter: %w", err)
+			log.Warnf("decrement set counter for stale rule %s: %v", ruleKey, err)
 		}
+		delete(r.rules, ruleKey)
+		return nil
+	}
+
+	if err := r.conn.DelRule(rule); err != nil {
+		return fmt.Errorf("remove legacy forwarding rule %s -> %s: %w", pair.Source, pair.Destination, err)
+	}
+
+	log.Debugf("removed legacy forwarding rule %s -> %s", pair.Source, pair.Destination)
+
+	delete(r.rules, ruleKey)
+
+	if err := r.decrementSetCounter(rule); err != nil {
+		return fmt.Errorf("decrement set counter: %w", err)
 	}
 
 	return nil
@@ -1006,17 +1070,22 @@ func (r *router) acceptFilterTableRules() error {
 		log.Debugf("Used %s to add accept forward and input rules", fw)
 	}()
 
-	// Try iptables first and fallback to nftables if iptables is not available
-	ipt, err := iptables.New()
+	// Try iptables first and fallback to nftables if iptables is not available.
+	// Use the correct protocol (iptables vs ip6tables) for the address family.
+	ipt, err := iptables.NewWithProtocol(r.iptablesProto())
 	if err != nil {
-		// iptables is not available but the filter table exists
 		log.Warnf("Will use nftables to manipulate the filter table because iptables is not available: %v", err)
 
 		fw = "nftables"
 		return r.acceptFilterRulesNftables(r.filterTable)
 	}
 
-	return r.acceptFilterRulesIptables(ipt)
+	if err := r.acceptFilterRulesIptables(ipt); err != nil {
+		log.Warnf("iptables failed (table may be incompatible), falling back to nftables: %v", err)
+		fw = "nftables"
+		return r.acceptFilterRulesNftables(r.filterTable)
+	}
+	return nil
 }
 
 func (r *router) acceptFilterRulesIptables(ipt *iptables.IPTables) error {
@@ -1185,13 +1254,17 @@ func (r *router) removeFilterTableRules() error {
 		return nil
 	}
 
-	ipt, err := iptables.New()
+	ipt, err := iptables.NewWithProtocol(r.iptablesProto())
 	if err != nil {
 		log.Debugf("iptables not available, using nftables to remove filter rules: %v", err)
 		return r.removeAcceptRulesFromTable(r.filterTable)
 	}
 
-	return r.removeAcceptFilterRulesIptables(ipt)
+	if err := r.removeAcceptFilterRulesIptables(ipt); err != nil {
+		log.Debugf("iptables removal failed (table may be incompatible), falling back to nftables: %v", err)
+		return r.removeAcceptRulesFromTable(r.filterTable)
+	}
+	return nil
 }
 
 func (r *router) removeAcceptRulesFromTable(table *nftables.Table) error {
@@ -1258,7 +1331,7 @@ func (r *router) removeExternalChainsRules() error {
 func (r *router) findExternalChains() []*nftables.Chain {
 	var chains []*nftables.Chain
 
-	families := []nftables.TableFamily{nftables.TableFamilyIPv4, nftables.TableFamilyINet}
+	families := []nftables.TableFamily{r.af.tableFamily, nftables.TableFamilyINet}
 
 	for _, family := range families {
 		allChains, err := r.conn.ListChainsOfTableFamily(family)
@@ -1282,8 +1355,8 @@ func (r *router) isExternalChain(chain *nftables.Chain) bool {
 		return false
 	}
 
-	// Skip all iptables-managed tables in the ip family
-	if chain.Table.Family == nftables.TableFamilyIPv4 && isIptablesTable(chain.Table.Name) {
+	// Skip iptables/ip6tables-managed tables (adding nft-native rules breaks iptables-save compat)
+	if (chain.Table.Family == nftables.TableFamilyIPv4 || chain.Table.Family == nftables.TableFamilyIPv6) && isIptablesTable(chain.Table.Name) {
 		return false
 	}
 
@@ -1329,65 +1402,89 @@ func (r *router) RemoveNatRule(pair firewall.RouterPair) error {
 		return fmt.Errorf(refreshRulesMapError, err)
 	}
 
+	var merr *multierror.Error
+
 	if pair.Masquerade {
 		if err := r.removeNatRule(pair); err != nil {
-			return fmt.Errorf("remove prerouting rule: %w", err)
+			merr = multierror.Append(merr, fmt.Errorf("remove prerouting rule: %w", err))
 		}
 
 		if err := r.removeNatRule(firewall.GetInversePair(pair)); err != nil {
-			return fmt.Errorf("remove inverse prerouting rule: %w", err)
+			merr = multierror.Append(merr, fmt.Errorf("remove inverse prerouting rule: %w", err))
 		}
 	}
 
 	if err := r.removeLegacyRouteRule(pair); err != nil {
-		return fmt.Errorf("remove legacy routing rule: %w", err)
+		merr = multierror.Append(merr, fmt.Errorf("remove legacy routing rule: %w", err))
 	}
 
+	// Set counters are decremented in the sub-methods above before flush. If flush fails,
+	// counters will be off until the next successful removal or refresh cycle.
 	if err := r.conn.Flush(); err != nil {
-		// TODO: rollback set counter
-		return fmt.Errorf("remove nat rules rule %s: %v", pair.Destination, err)
+		merr = multierror.Append(merr, fmt.Errorf("flush remove nat rules %s: %w", pair.Destination, err))
 	}
 
-	return nil
+	return nberrors.FormatErrorOrNil(merr)
 }
 
 func (r *router) removeNatRule(pair firewall.RouterPair) error {
 	ruleKey := firewall.GenKey(firewall.PreroutingFormat, pair)
 
-	if rule, exists := r.rules[ruleKey]; exists {
-		if err := r.conn.DelRule(rule); err != nil {
-			return fmt.Errorf("remove prerouting rule %s -> %s: %v", pair.Source, pair.Destination, err)
-		}
-
-		log.Debugf("removed prerouting rule %s -> %s", pair.Source, pair.Destination)
-
-		delete(r.rules, ruleKey)
-
-		if err := r.decrementSetCounter(rule); err != nil {
-			return fmt.Errorf("decrement set counter: %w", err)
-		}
-	} else {
+	rule, exists := r.rules[ruleKey]
+	if !exists {
 		log.Debugf("prerouting rule %s not found", ruleKey)
+		return nil
+	}
+
+	if rule.Handle == 0 {
+		log.Warnf("prerouting rule %s has no handle, removing stale entry", ruleKey)
+		if err := r.decrementSetCounter(rule); err != nil {
+			log.Warnf("decrement set counter for stale rule %s: %v", ruleKey, err)
+		}
+		delete(r.rules, ruleKey)
+		return nil
+	}
+
+	if err := r.conn.DelRule(rule); err != nil {
+		return fmt.Errorf("remove prerouting rule %s -> %s: %w", pair.Source, pair.Destination, err)
+	}
+
+	log.Debugf("removed prerouting rule %s -> %s", pair.Source, pair.Destination)
+
+	delete(r.rules, ruleKey)
+
+	if err := r.decrementSetCounter(rule); err != nil {
+		return fmt.Errorf("decrement set counter: %w", err)
 	}
 
 	return nil
 }
 
-// refreshRulesMap refreshes the rule map with the latest rules. this is useful to avoid
-// duplicates and to get missing attributes that we don't have when adding new rules
+// refreshRulesMap rebuilds the rule map from the kernel. This removes stale entries
+// (e.g. from failed flushes) and updates handles for all existing rules.
 func (r *router) refreshRulesMap() error {
+	var merr *multierror.Error
+	newRules := make(map[string]*nftables.Rule)
 	for _, chain := range r.chains {
 		rules, err := r.conn.GetRules(chain.Table, chain)
 		if err != nil {
-			return fmt.Errorf("list rules: %w", err)
+			merr = multierror.Append(merr, fmt.Errorf("list rules for chain %s: %w", chain.Name, err))
+			// preserve existing entries for this chain since we can't verify their state
+			for k, v := range r.rules {
+				if v.Chain != nil && v.Chain.Name == chain.Name {
+					newRules[k] = v
+				}
+			}
+			continue
 		}
 		for _, rule := range rules {
 			if len(rule.UserData) > 0 {
-				r.rules[string(rule.UserData)] = rule
+				newRules[string(rule.UserData)] = rule
 			}
 		}
 	}
-	return nil
+	r.rules = newRules
+	return nberrors.FormatErrorOrNil(merr)
 }
 
 func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
@@ -1400,7 +1497,7 @@ func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
 		return rule, nil
 	}
 
-	protoNum, err := protoToInt(rule.Protocol)
+	protoNum, err := r.af.protoNum(rule.Protocol)
 	if err != nil {
 		return nil, fmt.Errorf("convert protocol to number: %w", err)
 	}
@@ -1463,7 +1560,7 @@ func (r *router) addDnatRedirect(rule firewall.ForwardRule, protoNum uint8, rule
 	dnatExprs = append(dnatExprs,
 		&expr.NAT{
 			Type:        expr.NATTypeDestNAT,
-			Family:      uint32(nftables.TableFamilyIPv4),
+			Family:      uint32(r.af.tableFamily),
 			RegAddrMin:  1,
 			RegProtoMin: regProtoMin,
 			RegProtoMax: regProtoMax,
@@ -1559,7 +1656,7 @@ func (r *router) addXTablesRedirect(dnatExprs []expr.Any, ruleKey string, rule f
 	dnatRule := &nftables.Rule{
 		Table: &nftables.Table{
 			Name:   tableNat,
-			Family: nftables.TableFamilyIPv4,
+			Family: r.af.tableFamily,
 		},
 		Chain: &nftables.Chain{
 			Name:     chainNameNatPrerouting,
@@ -1594,8 +1691,8 @@ func (r *router) addDnatMasq(rule firewall.ForwardRule, protoNum uint8, ruleKey
 		&expr.Payload{
 			DestRegister: 1,
 			Base:         expr.PayloadBaseNetworkHeader,
-			Offset:       16,
-			Len:          4,
+			Offset:       r.af.dstAddrOffset,
+			Len:          r.af.addrLen,
 		},
 		&expr.Cmp{
 			Op:       expr.CmpOpEq,
@@ -1629,20 +1726,34 @@ func (r *router) DeleteDNATRule(rule firewall.Rule) error {
 	}
 
 	var merr *multierror.Error
+	var needsFlush bool
+
 	if dnatRule, exists := r.rules[ruleKey+dnatSuffix]; exists {
-		if err := r.conn.DelRule(dnatRule); err != nil {
+		if dnatRule.Handle == 0 {
+			log.Warnf("dnat rule %s has no handle, removing stale entry", ruleKey+dnatSuffix)
+			delete(r.rules, ruleKey+dnatSuffix)
+		} else if err := r.conn.DelRule(dnatRule); err != nil {
 			merr = multierror.Append(merr, fmt.Errorf("delete dnat rule: %w", err))
+		} else {
+			needsFlush = true
 		}
 	}
 
 	if masqRule, exists := r.rules[ruleKey+snatSuffix]; exists {
-		if err := r.conn.DelRule(masqRule); err != nil {
+		if masqRule.Handle == 0 {
+			log.Warnf("snat rule %s has no handle, removing stale entry", ruleKey+snatSuffix)
+			delete(r.rules, ruleKey+snatSuffix)
+		} else if err := r.conn.DelRule(masqRule); err != nil {
 			merr = multierror.Append(merr, fmt.Errorf("delete snat rule: %w", err))
+		} else {
+			needsFlush = true
 		}
 	}
 
-	if err := r.conn.Flush(); err != nil {
-		merr = multierror.Append(merr, fmt.Errorf(flushError, err))
+	if needsFlush {
+		if err := r.conn.Flush(); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf(flushError, err))
+		}
 	}
 
 	if merr == nil {
@@ -1659,7 +1770,7 @@ func (r *router) UpdateSet(set firewall.Set, prefixes []netip.Prefix) error {
 		return fmt.Errorf("get set %s: %w", set.HashedName(), err)
 	}
 
-	elements := convertPrefixesToSet(prefixes)
+	elements := r.convertPrefixesToSet(prefixes)
 	if err := r.conn.SetAddElements(nfset, elements); err != nil {
 		return fmt.Errorf("add elements to set %s: %w", set.HashedName(), err)
 	}
@@ -1681,7 +1792,7 @@ func (r *router) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol
 		return nil
 	}
 
-	protoNum, err := protoToInt(protocol)
+	protoNum, err := r.af.protoNum(protocol)
 	if err != nil {
 		return fmt.Errorf("convert protocol to number: %w", err)
 	}
@@ -1712,7 +1823,11 @@ func (r *router) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol
 		},
 	}
 
-	exprs = append(exprs, applyPrefix(netip.PrefixFrom(localAddr, 32), false)...)
+	bits := 32
+	if localAddr.Is6() {
+		bits = 128
+	}
+	exprs = append(exprs, r.applyPrefix(netip.PrefixFrom(localAddr, bits), false)...)
 
 	exprs = append(exprs,
 		&expr.Immediate{
@@ -1725,7 +1840,7 @@ func (r *router) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol
 		},
 		&expr.NAT{
 			Type:        expr.NATTypeDestNAT,
-			Family:      uint32(nftables.TableFamilyIPv4),
+			Family:      uint32(r.af.tableFamily),
 			RegAddrMin:  1,
 			RegProtoMin: 2,
 			RegProtoMax: 0,
@@ -1757,16 +1872,25 @@ func (r *router) RemoveInboundDNAT(localAddr netip.Addr, protocol firewall.Proto
 
 	ruleID := fmt.Sprintf("inbound-dnat-%s-%s-%d-%d", localAddr.String(), protocol, sourcePort, targetPort)
 
-	if rule, exists := r.rules[ruleID]; exists {
-		if err := r.conn.DelRule(rule); err != nil {
-			return fmt.Errorf("delete inbound DNAT rule %s: %w", ruleID, err)
-		}
-		if err := r.conn.Flush(); err != nil {
-			return fmt.Errorf("flush delete inbound DNAT rule: %w", err)
-		}
-		delete(r.rules, ruleID)
+	rule, exists := r.rules[ruleID]
+	if !exists {
+		return nil
 	}
 
+	if rule.Handle == 0 {
+		log.Warnf("inbound DNAT rule %s has no handle, removing stale entry", ruleID)
+		delete(r.rules, ruleID)
+		return nil
+	}
+
+	if err := r.conn.DelRule(rule); err != nil {
+		return fmt.Errorf("delete inbound DNAT rule %s: %w", ruleID, err)
+	}
+	if err := r.conn.Flush(); err != nil {
+		return fmt.Errorf("flush delete inbound DNAT rule: %w", err)
+	}
+	delete(r.rules, ruleID)
+
 	return nil
 }
 
@@ -1785,45 +1909,44 @@ func (r *router) applyNetwork(
 	}
 
 	if network.IsPrefix() {
-		return applyPrefix(network.Prefix, isSource), nil
+		return r.applyPrefix(network.Prefix, isSource), nil
 	}
 
 	return nil, nil
 }
 
 // applyPrefix generates nftables expressions for a CIDR prefix
-func applyPrefix(prefix netip.Prefix, isSource bool) []expr.Any {
-	// dst offset
-	offset := uint32(16)
+func (r *router) applyPrefix(prefix netip.Prefix, isSource bool) []expr.Any {
+	// dst offset by default
+	offset := r.af.dstAddrOffset
 	if isSource {
 		// src offset
-		offset = 12
+		offset = r.af.srcAddrOffset
 	}
 
 	ones := prefix.Bits()
-	// 0.0.0.0/0 doesn't need extra expressions
+	// unspecified address (/0) doesn't need extra expressions
 	if ones == 0 {
 		return nil
 	}
 
-	mask := net.CIDRMask(ones, 32)
+	mask := net.CIDRMask(ones, r.af.totalBits)
+	xor := make([]byte, r.af.addrLen)
 
 	return []expr.Any{
 		&expr.Payload{
 			DestRegister: 1,
 			Base:         expr.PayloadBaseNetworkHeader,
 			Offset:       offset,
-			Len:          4,
+			Len:          r.af.addrLen,
 		},
-		// netmask
 		&expr.Bitwise{
 			DestRegister:   1,
 			SourceRegister: 1,
-			Len:            4,
+			Len:            r.af.addrLen,
 			Mask:           mask,
-			Xor:            []byte{0, 0, 0, 0},
+			Xor:            xor,
 		},
-		// net address
 		&expr.Cmp{
 			Op:       expr.CmpOpEq,
 			Register: 1,
@@ -1906,13 +2029,12 @@ func getCtNewExprs() []expr.Any {
 	}
 }
 
-func getIpSetExprs(ref refcounter.Ref[*nftables.Set], isSource bool) ([]expr.Any, error) {
-
-	// dst offset
-	offset := uint32(16)
+func (r *router) getIpSetExprs(ref refcounter.Ref[*nftables.Set], isSource bool) ([]expr.Any, error) {
+	// dst offset by default
+	offset := r.af.dstAddrOffset
 	if isSource {
 		// src offset
-		offset = 12
+		offset = r.af.srcAddrOffset
 	}
 
 	return []expr.Any{
@@ -1920,7 +2042,7 @@ func getIpSetExprs(ref refcounter.Ref[*nftables.Set], isSource bool) ([]expr.Any
 			DestRegister: 1,
 			Base:         expr.PayloadBaseNetworkHeader,
 			Offset:       offset,
-			Len:          4,
+			Len:          r.af.addrLen,
 		},
 		&expr.Lookup{
 			SourceRegister: 1,

client/firewall/nftables/router_linux_test.go

@@ -18,6 +18,7 @@ import (
 	firewall "github.com/netbirdio/netbird/client/firewall/manager"
 	"github.com/netbirdio/netbird/client/firewall/test"
 	"github.com/netbirdio/netbird/client/iface"
+	"github.com/netbirdio/netbird/client/internal/acl/id"
 )
 
 const (
@@ -89,8 +90,9 @@ func TestNftablesManager_AddNatRule(t *testing.T) {
 				}
 
 				// Build CIDR matching expressions
-				sourceExp := applyPrefix(testCase.InputPair.Source.Prefix, true)
-				destExp := applyPrefix(testCase.InputPair.Destination.Prefix, false)
+				testRouter := &router{af: afIPv4}
+				sourceExp := testRouter.applyPrefix(testCase.InputPair.Source.Prefix, true)
+				destExp := testRouter.applyPrefix(testCase.InputPair.Destination.Prefix, false)
 
 				// Combine all expressions in the correct order
 				// nolint:gocritic
@@ -507,6 +509,136 @@ func TestNftablesCreateIpSet(t *testing.T) {
 	}
 }
 
+func TestNftablesCreateIpSet_IPv6(t *testing.T) {
+	if check() != NFTABLES {
+		t.Skip("nftables not supported on this system")
+	}
+
+	workTable, err := createWorkTableIPv6()
+	require.NoError(t, err, "Failed to create v6 work table")
+	defer deleteWorkTableIPv6()
+
+	r, err := newRouter(workTable, ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err, "Failed to create router")
+	require.NoError(t, r.init(workTable))
+	defer func() {
+		require.NoError(t, r.Reset(), "Failed to reset router")
+	}()
+
+	tests := []struct {
+		name     string
+		sources  []netip.Prefix
+		expected []netip.Prefix
+	}{
+		{
+			name:    "Single IPv6",
+			sources: []netip.Prefix{netip.MustParsePrefix("2001:db8::1/128")},
+		},
+		{
+			name: "Multiple IPv6 Subnets",
+			sources: []netip.Prefix{
+				netip.MustParsePrefix("fd00::/64"),
+				netip.MustParsePrefix("2001:db8::/48"),
+				netip.MustParsePrefix("fe80::/10"),
+			},
+		},
+		{
+			name: "Overlapping IPv6",
+			sources: []netip.Prefix{
+				netip.MustParsePrefix("fd00::/48"),
+				netip.MustParsePrefix("fd00::/64"),
+				netip.MustParsePrefix("fd00::1/128"),
+			},
+			expected: []netip.Prefix{
+				netip.MustParsePrefix("fd00::/48"),
+			},
+		},
+		{
+			name: "Mixed prefix lengths",
+			sources: []netip.Prefix{
+				netip.MustParsePrefix("2001:db8:1::/48"),
+				netip.MustParsePrefix("2001:db8:2::1/128"),
+				netip.MustParsePrefix("fd00:abcd::/32"),
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			setName := firewall.NewPrefixSet(tt.sources).HashedName()
+			set, err := r.createIpSet(setName, setInput{prefixes: tt.sources})
+			require.NoError(t, err, "Failed to create IPv6 set")
+			require.NotNil(t, set)
+
+			assert.Equal(t, setName, set.Name)
+			assert.True(t, set.Interval)
+			assert.Equal(t, nftables.TypeIP6Addr, set.KeyType)
+
+			fetchedSet, err := r.conn.GetSetByName(r.workTable, setName)
+			require.NoError(t, err, "Failed to fetch created set")
+
+			elements, err := r.conn.GetSetElements(fetchedSet)
+			require.NoError(t, err, "Failed to get set elements")
+
+			uniquePrefixes := make(map[string]bool)
+			for _, elem := range elements {
+				if !elem.IntervalEnd && len(elem.Key) == 16 {
+					ip := netip.AddrFrom16([16]byte(elem.Key))
+					uniquePrefixes[ip.String()] = true
+				}
+			}
+
+			expectedCount := len(tt.expected)
+			if expectedCount == 0 {
+				expectedCount = len(tt.sources)
+			}
+			assert.Equal(t, expectedCount, len(uniquePrefixes), "unique prefix count mismatch")
+
+			r.conn.DelSet(set)
+			require.NoError(t, r.conn.Flush())
+		})
+	}
+}
+
+func createWorkTableIPv6() (*nftables.Table, error) {
+	sConn, err := nftables.New(nftables.AsLasting())
+	if err != nil {
+		return nil, err
+	}
+
+	tables, err := sConn.ListTablesOfFamily(nftables.TableFamilyIPv6)
+	if err != nil {
+		return nil, err
+	}
+	for _, t := range tables {
+		if t.Name == tableNameNetbird {
+			sConn.DelTable(t)
+		}
+	}
+
+	table := sConn.AddTable(&nftables.Table{Name: tableNameNetbird, Family: nftables.TableFamilyIPv6})
+	err = sConn.Flush()
+	return table, err
+}
+
+func deleteWorkTableIPv6() {
+	sConn, err := nftables.New(nftables.AsLasting())
+	if err != nil {
+		return
+	}
+
+	tables, err := sConn.ListTablesOfFamily(nftables.TableFamilyIPv6)
+	if err != nil {
+		return
+	}
+	for _, t := range tables {
+		if t.Name == tableNameNetbird {
+			sConn.DelTable(t)
+			_ = sConn.Flush()
+		}
+	}
+}
+
 func verifyRule(t *testing.T, rule *nftables.Rule, sources []netip.Prefix, destination netip.Prefix, proto firewall.Protocol, sPort, dPort *firewall.Port, direction firewall.RuleDirection, action firewall.Action, expectSet bool) {
 	t.Helper()
 
@@ -626,7 +758,7 @@ func containsPort(exprs []expr.Any, port *firewall.Port, isSource bool) bool {
 
 func containsProtocol(exprs []expr.Any, proto firewall.Protocol) bool {
 	var metaFound, cmpFound bool
-	expectedProto, _ := protoToInt(proto)
+	expectedProto, _ := afIPv4.protoNum(proto)
 	for _, e := range exprs {
 		switch ex := e.(type) {
 		case *expr.Meta:
@@ -719,3 +851,189 @@ func deleteWorkTable() {
 		}
 	}
 }
+
+func TestRouter_RefreshRulesMap_RemovesStaleEntries(t *testing.T) {
+	if check() != NFTABLES {
+		t.Skip("nftables not supported on this system")
+	}
+
+	workTable, err := createWorkTable()
+	require.NoError(t, err)
+	defer deleteWorkTable()
+
+	r, err := newRouter(workTable, ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, r.init(workTable))
+	defer func() { require.NoError(t, r.Reset()) }()
+
+	// Add a real rule to the kernel
+	ruleKey, err := r.AddRouteFiltering(
+		nil,
+		[]netip.Prefix{netip.MustParsePrefix("192.168.1.0/24")},
+		firewall.Network{Prefix: netip.MustParsePrefix("10.0.0.0/24")},
+		firewall.ProtocolTCP,
+		nil,
+		&firewall.Port{Values: []uint16{80}},
+		firewall.ActionAccept,
+	)
+	require.NoError(t, err)
+	t.Cleanup(func() {
+		require.NoError(t, r.DeleteRouteRule(ruleKey))
+	})
+
+	// Inject a stale entry with Handle=0 (simulates store-before-flush failure)
+	staleKey := "stale-rule-that-does-not-exist"
+	r.rules[staleKey] = &nftables.Rule{
+		Table:    r.workTable,
+		Chain:    r.chains[chainNameRoutingFw],
+		Handle:   0,
+		UserData: []byte(staleKey),
+	}
+
+	require.Contains(t, r.rules, staleKey, "stale entry should be in map before refresh")
+
+	err = r.refreshRulesMap()
+	require.NoError(t, err)
+
+	assert.NotContains(t, r.rules, staleKey, "stale entry should be removed after refresh")
+
+	realRule, ok := r.rules[ruleKey.ID()]
+	assert.True(t, ok, "real rule should still exist after refresh")
+	assert.NotZero(t, realRule.Handle, "real rule should have a valid handle")
+}
+
+func TestRouter_DeleteRouteRule_StaleHandle(t *testing.T) {
+	if check() != NFTABLES {
+		t.Skip("nftables not supported on this system")
+	}
+
+	workTable, err := createWorkTable()
+	require.NoError(t, err)
+	defer deleteWorkTable()
+
+	r, err := newRouter(workTable, ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, r.init(workTable))
+	defer func() { require.NoError(t, r.Reset()) }()
+
+	// Inject a stale entry with Handle=0
+	staleKey := "stale-route-rule"
+	r.rules[staleKey] = &nftables.Rule{
+		Table:    r.workTable,
+		Chain:    r.chains[chainNameRoutingFw],
+		Handle:   0,
+		UserData: []byte(staleKey),
+	}
+
+	// DeleteRouteRule should not return an error for stale handles
+	err = r.DeleteRouteRule(id.RuleID(staleKey))
+	assert.NoError(t, err, "deleting a stale rule should not error")
+	assert.NotContains(t, r.rules, staleKey, "stale entry should be cleaned up")
+}
+
+func TestRouter_AddNatRule_WithStaleEntry(t *testing.T) {
+	if check() != NFTABLES {
+		t.Skip("nftables not supported on this system")
+	}
+
+	manager, err := Create(ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, manager.Init(nil))
+	t.Cleanup(func() {
+		require.NoError(t, manager.Close(nil))
+	})
+
+	pair := firewall.RouterPair{
+		ID:          "staletest",
+		Source:      firewall.Network{Prefix: netip.MustParsePrefix("100.100.100.1/32")},
+		Destination: firewall.Network{Prefix: netip.MustParsePrefix("100.100.200.0/24")},
+		Masquerade:  true,
+	}
+
+	rtr := manager.router
+
+	// First add succeeds
+	err = rtr.AddNatRule(pair)
+	require.NoError(t, err)
+	t.Cleanup(func() {
+		require.NoError(t, rtr.RemoveNatRule(pair))
+	})
+
+	// Corrupt the handle to simulate stale state
+	natRuleKey := firewall.GenKey(firewall.PreroutingFormat, pair)
+	if rule, exists := rtr.rules[natRuleKey]; exists {
+		rule.Handle = 0
+	}
+	inverseKey := firewall.GenKey(firewall.PreroutingFormat, firewall.GetInversePair(pair))
+	if rule, exists := rtr.rules[inverseKey]; exists {
+		rule.Handle = 0
+	}
+
+	// Adding the same rule again should succeed despite stale handles
+	err = rtr.AddNatRule(pair)
+	assert.NoError(t, err, "AddNatRule should succeed even with stale entries")
+
+	// Verify rules exist in kernel
+	rules, err := rtr.conn.GetRules(rtr.workTable, rtr.chains[chainNameManglePrerouting])
+	require.NoError(t, err)
+
+	found := 0
+	for _, rule := range rules {
+		if len(rule.UserData) > 0 && string(rule.UserData) == natRuleKey {
+			found++
+		}
+	}
+	assert.Equal(t, 1, found, "NAT rule should exist in kernel")
+}
+
+func TestCalculateLastIP(t *testing.T) {
+	tests := []struct {
+		prefix string
+		want   string
+	}{
+		{"10.0.0.0/24", "10.0.0.255"},
+		{"10.0.0.0/32", "10.0.0.0"},
+		{"0.0.0.0/0", "255.255.255.255"},
+		{"192.168.1.0/28", "192.168.1.15"},
+		{"fd00::/64", "fd00::ffff:ffff:ffff:ffff"},
+		{"fd00::/128", "fd00::"},
+		{"2001:db8::/48", "2001:db8:0:ffff:ffff:ffff:ffff:ffff"},
+		{"::/0", "ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff"},
+	}
+	for _, tt := range tests {
+		t.Run(tt.prefix, func(t *testing.T) {
+			prefix := netip.MustParsePrefix(tt.prefix)
+			got := calculateLastIP(prefix)
+			assert.Equal(t, tt.want, got.String())
+		})
+	}
+}
+
+func TestConvertPrefixesToSet_IPv6(t *testing.T) {
+	r := &router{af: afIPv6}
+	prefixes := []netip.Prefix{
+		netip.MustParsePrefix("fd00::/64"),
+		netip.MustParsePrefix("2001:db8::1/128"),
+	}
+
+	elements := r.convertPrefixesToSet(prefixes)
+
+	// Each prefix produces 2 elements (start + end)
+	require.Len(t, elements, 4)
+
+	// fd00::/64 start
+	assert.Equal(t, netip.MustParseAddr("fd00::").As16(), [16]byte(elements[0].Key))
+	assert.False(t, elements[0].IntervalEnd)
+
+	// fd00::/64 end (fd00:0:0:1::, one past the last)
+	assert.Equal(t, netip.MustParseAddr("fd00:0:0:1::").As16(), [16]byte(elements[1].Key))
+	assert.True(t, elements[1].IntervalEnd)
+
+	// 2001:db8::1/128 start
+	assert.Equal(t, netip.MustParseAddr("2001:db8::1").As16(), [16]byte(elements[2].Key))
+	assert.False(t, elements[2].IntervalEnd)
+
+	// 2001:db8::1/128 end (2001:db8::2)
+	assert.Equal(t, netip.MustParseAddr("2001:db8::2").As16(), [16]byte(elements[3].Key))
+	assert.True(t, elements[3].IntervalEnd)
+}

client/firewall/uspfilter/allow_netbird.go

@@ -3,12 +3,6 @@
 package uspfilter
 
 import (
-	"context"
-	"net/netip"
-	"time"
-
-	log "github.com/sirupsen/logrus"
-
 	"github.com/netbirdio/netbird/client/internal/statemanager"
 )
 
@@ -17,33 +11,7 @@ func (m *Manager) Close(stateManager *statemanager.Manager) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	m.outgoingRules = make(map[netip.Addr]RuleSet)
-	m.incomingDenyRules = make(map[netip.Addr]RuleSet)
-	m.incomingRules = make(map[netip.Addr]RuleSet)
-
-	if m.udpTracker != nil {
-		m.udpTracker.Close()
-	}
-
-	if m.icmpTracker != nil {
-		m.icmpTracker.Close()
-	}
-
-	if m.tcpTracker != nil {
-		m.tcpTracker.Close()
-	}
-
-	if fwder := m.forwarder.Load(); fwder != nil {
-		fwder.Stop()
-	}
-
-	if m.logger != nil {
-		ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
-		defer cancel()
-		if err := m.logger.Stop(ctx); err != nil {
-			log.Errorf("failed to shutdown logger: %v", err)
-		}
-	}
+	m.resetState()
 
 	if m.nativeFirewall != nil {
 		return m.nativeFirewall.Close(stateManager)

client/firewall/uspfilter/allow_netbird_windows.go

@@ -1,15 +1,14 @@
 package uspfilter
 
 import (
-	"context"
 	"fmt"
-	"net/netip"
 	"os/exec"
 	"syscall"
-	"time"
 
+	"github.com/hashicorp/go-multierror"
 	log "github.com/sirupsen/logrus"
 
+	nberrors "github.com/netbirdio/netbird/client/errors"
 	"github.com/netbirdio/netbird/client/internal/statemanager"
 )
 
@@ -26,47 +25,26 @@ func (m *Manager) Close(*statemanager.Manager) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	m.outgoingRules = make(map[netip.Addr]RuleSet)
-	m.incomingDenyRules = make(map[netip.Addr]RuleSet)
-	m.incomingRules = make(map[netip.Addr]RuleSet)
-
-	if m.udpTracker != nil {
-		m.udpTracker.Close()
-	}
-
-	if m.icmpTracker != nil {
-		m.icmpTracker.Close()
-	}
-
-	if m.tcpTracker != nil {
-		m.tcpTracker.Close()
-	}
-
-	if fwder := m.forwarder.Load(); fwder != nil {
-		fwder.Stop()
-	}
-
-	if m.logger != nil {
-		ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
-		defer cancel()
-		if err := m.logger.Stop(ctx); err != nil {
-			log.Errorf("failed to shutdown logger: %v", err)
-		}
-	}
+	m.resetState()
 
 	if !isWindowsFirewallReachable() {
 		return nil
 	}
 
-	if !isFirewallRuleActive(firewallRuleName) {
-		return nil
+	var merr *multierror.Error
+	if isFirewallRuleActive(firewallRuleName) {
+		if err := manageFirewallRule(firewallRuleName, deleteRule); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove windows firewall rule: %w", err))
+		}
 	}
 
-	if err := manageFirewallRule(firewallRuleName, deleteRule); err != nil {
-		return fmt.Errorf("couldn't remove windows firewall: %w", err)
+	if isFirewallRuleActive(firewallRuleName + "-v6") {
+		if err := manageFirewallRule(firewallRuleName+"-v6", deleteRule); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove windows v6 firewall rule: %w", err))
+		}
 	}
 
-	return nil
+	return nberrors.FormatErrorOrNil(merr)
 }
 
 // AllowNetbird allows netbird interface traffic
@@ -75,17 +53,33 @@ func (m *Manager) AllowNetbird() error {
 		return nil
 	}
 
-	if isFirewallRuleActive(firewallRuleName) {
-		return nil
+	if !isFirewallRuleActive(firewallRuleName) {
+		if err := manageFirewallRule(firewallRuleName,
+			addRule,
+			"dir=in",
+			"enable=yes",
+			"action=allow",
+			"profile=any",
+			"localip="+m.wgIface.Address().IP.String(),
+		); err != nil {
+			return err
+		}
 	}
-	return manageFirewallRule(firewallRuleName,
-		addRule,
-		"dir=in",
-		"enable=yes",
-		"action=allow",
-		"profile=any",
-		"localip="+m.wgIface.Address().IP.String(),
-	)
+
+	if v6 := m.wgIface.Address().IPv6; v6.IsValid() && !isFirewallRuleActive(firewallRuleName+"-v6") {
+		if err := manageFirewallRule(firewallRuleName+"-v6",
+			addRule,
+			"dir=in",
+			"enable=yes",
+			"action=allow",
+			"profile=any",
+			"localip="+v6.String(),
+		); err != nil {
+			return err
+		}
+	}
+
+	return nil
 }
 
 func manageFirewallRule(ruleName string, action action, extraArgs ...string) error {

client/firewall/uspfilter/conntrack/common.go

@@ -1,8 +1,9 @@
 package conntrack
 
 import (
-	"fmt"
+	"net"
 	"net/netip"
+	"strconv"
 	"sync/atomic"
 	"time"
 
@@ -64,5 +65,7 @@ type ConnKey struct {
 }
 
 func (c ConnKey) String() string {
-	return fmt.Sprintf("%s:%d → %s:%d", c.SrcIP.Unmap(), c.SrcPort, c.DstIP.Unmap(), c.DstPort)
+	return net.JoinHostPort(c.SrcIP.Unmap().String(), strconv.Itoa(int(c.SrcPort))) +
+		" → " +
+		net.JoinHostPort(c.DstIP.Unmap().String(), strconv.Itoa(int(c.DstPort)))
 }

client/firewall/uspfilter/conntrack/icmp.go

@@ -21,9 +21,10 @@ const (
 	// ICMPCleanupInterval is how often we check for stale ICMP connections
 	ICMPCleanupInterval = 15 * time.Second
 
-	// MaxICMPPayloadLength is the maximum length of ICMP payload we consider for original packet info,
-	// which includes the IP header (20 bytes) and transport header (8 bytes)
-	MaxICMPPayloadLength = 28
+	// MaxICMPPayloadLength is the maximum length of ICMP payload we consider for original packet info.
+	// IPv4: 20-byte header + 8-byte transport = 28 bytes.
+	// IPv6: 40-byte header + 8-byte transport = 48 bytes.
+	MaxICMPPayloadLength = 48
 )
 
 // ICMPConnKey uniquely identifies an ICMP connection
@@ -74,32 +75,64 @@ func (info ICMPInfo) String() string {
 	return info.TypeCode.String()
 }
 
-// isErrorMessage returns true if this ICMP type carries original packet info
+// isErrorMessage returns true if this ICMP type carries original packet info.
+// Covers both ICMPv4 and ICMPv6 error types. Without a family field we match
+// both sets; type 3 overlaps (v4 DestUnreachable / v6 TimeExceeded) so it's
+// kept as a literal.
 func (info ICMPInfo) isErrorMessage() bool {
 	typ := info.TypeCode.Type()
-	return typ == 3 || // Destination Unreachable
-		typ == 5 || // Redirect
-		typ == 11 || // Time Exceeded
-		typ == 12 // Parameter Problem
+	// ICMPv4 error types
+	if typ == layers.ICMPv4TypeDestinationUnreachable ||
+		typ == layers.ICMPv4TypeRedirect ||
+		typ == layers.ICMPv4TypeTimeExceeded ||
+		typ == layers.ICMPv4TypeParameterProblem {
+		return true
+	}
+	// ICMPv6 error types (type 3 already matched above as v4 DestUnreachable)
+	if typ == layers.ICMPv6TypeDestinationUnreachable ||
+		typ == layers.ICMPv6TypePacketTooBig ||
+		typ == layers.ICMPv6TypeParameterProblem {
+		return true
+	}
+	return false
 }
 
 // parseOriginalPacket extracts info about the original packet from ICMP payload
 func (info ICMPInfo) parseOriginalPacket() string {
-	if info.PayloadLen < MaxICMPPayloadLength {
+	if info.PayloadLen == 0 {
 		return ""
 	}
 
-	// TODO: handle IPv6
-	if version := (info.PayloadData[0] >> 4) & 0xF; version != 4 {
+	version := (info.PayloadData[0] >> 4) & 0xF
+
+	var protocol uint8
+	var srcIP, dstIP net.IP
+	var transportData []byte
+
+	switch version {
+	case 4:
+		// 20-byte IPv4 header + 8-byte transport minimum
+		if info.PayloadLen < 28 {
+			return ""
+		}
+		protocol = info.PayloadData[9]
+		srcIP = net.IP(info.PayloadData[12:16])
+		dstIP = net.IP(info.PayloadData[16:20])
+		transportData = info.PayloadData[20:]
+	case 6:
+		// 40-byte IPv6 header + 8-byte transport minimum
+		if info.PayloadLen < 48 {
+			return ""
+		}
+		// Next Header field in IPv6 header
+		protocol = info.PayloadData[6]
+		srcIP = net.IP(info.PayloadData[8:24])
+		dstIP = net.IP(info.PayloadData[24:40])
+		transportData = info.PayloadData[40:]
+	default:
 		return ""
 	}
 
-	protocol := info.PayloadData[9]
-	srcIP := net.IP(info.PayloadData[12:16])
-	dstIP := net.IP(info.PayloadData[16:20])
-
-	transportData := info.PayloadData[20:]
-
 	switch nftypes.Protocol(protocol) {
 	case nftypes.TCP:
 		srcPort := uint16(transportData[0])<<8 | uint16(transportData[1])
@@ -247,9 +280,10 @@ func (t *ICMPTracker) track(
 	t.sendEvent(nftypes.TypeStart, conn, ruleId)
 }
 
-// IsValidInbound checks if an inbound ICMP Echo Reply matches a tracked request
+// IsValidInbound checks if an inbound ICMP Echo Reply matches a tracked request.
+// Accepts both ICMPv4 (type 0) and ICMPv6 (type 129) echo replies.
 func (t *ICMPTracker) IsValidInbound(srcIP netip.Addr, dstIP netip.Addr, id uint16, icmpType uint8, size int) bool {
-	if icmpType != uint8(layers.ICMPv4TypeEchoReply) {
+	if icmpType != uint8(layers.ICMPv4TypeEchoReply) && icmpType != uint8(layers.ICMPv6TypeEchoReply) {
 		return false
 	}
 
@@ -301,6 +335,13 @@ func (t *ICMPTracker) cleanup() {
 	}
 }
 
+func icmpProtocolForAddr(ip netip.Addr) nftypes.Protocol {
+	if ip.Is6() {
+		return nftypes.ICMPv6
+	}
+	return nftypes.ICMP
+}
+
 // Close stops the cleanup routine and releases resources
 func (t *ICMPTracker) Close() {
 	t.tickerCancel()
@@ -316,7 +357,7 @@ func (t *ICMPTracker) sendEvent(typ nftypes.Type, conn *ICMPConnTrack, ruleID []
 		Type:      typ,
 		RuleID:    ruleID,
 		Direction: conn.Direction,
-		Protocol:  nftypes.ICMP, // TODO: adjust for IPv6/icmpv6
+		Protocol:  icmpProtocolForAddr(conn.SourceIP),
 		SourceIP:  conn.SourceIP,
 		DestIP:    conn.DestIP,
 		ICMPType:  conn.ICMPType,
@@ -334,7 +375,7 @@ func (t *ICMPTracker) sendStartEvent(direction nftypes.Direction, srcIP netip.Ad
 		Type:      nftypes.TypeStart,
 		RuleID:    ruleID,
 		Direction: direction,
-		Protocol:  nftypes.ICMP,
+		Protocol:  icmpProtocolForAddr(srcIP),
 		SourceIP:  srcIP,
 		DestIP:    dstIP,
 		ICMPType:  typ,

client/firewall/uspfilter/conntrack/tcp.go

@@ -115,6 +115,17 @@ func (t *TCPConnTrack) IsTombstone() bool {
 	return t.tombstone.Load()
 }
 
+// IsSupersededBy returns true if this connection should be replaced by a new one
+// carrying the given flags. Tombstoned connections are always superseded; TIME-WAIT
+// connections are superseded by a pure SYN (a new connection attempt for the same
+// four-tuple, as contemplated by RFC 1122 §4.2.2.13 and RFC 6191).
+func (t *TCPConnTrack) IsSupersededBy(flags uint8) bool {
+	if t.tombstone.Load() {
+		return true
+	}
+	return flags&TCPSyn != 0 && flags&TCPAck == 0 && TCPState(t.state.Load()) == TCPStateTimeWait
+}
+
 // SetTombstone safely marks the connection for deletion
 func (t *TCPConnTrack) SetTombstone() {
 	t.tombstone.Store(true)
@@ -169,7 +180,7 @@ func (t *TCPTracker) updateIfExists(srcIP, dstIP netip.Addr, srcPort, dstPort ui
 	conn, exists := t.connections[key]
 	t.mutex.RUnlock()
 
-	if exists {
+	if exists && !conn.IsSupersededBy(flags) {
 		t.updateState(key, conn, flags, direction, size)
 		return key, uint16(conn.DNATOrigPort.Load()), true
 	}
@@ -241,7 +252,7 @@ func (t *TCPTracker) IsValidInbound(srcIP, dstIP netip.Addr, srcPort, dstPort ui
 	conn, exists := t.connections[key]
 	t.mutex.RUnlock()
 
-	if !exists || conn.IsTombstone() {
+	if !exists || conn.IsSupersededBy(flags) {
 		return false
 	}
 

client/firewall/uspfilter/conntrack/tcp_test.go

@@ -485,6 +485,261 @@ func TestTCPAbnormalSequences(t *testing.T) {
 	})
 }
 
+// TestTCPPortReuseTombstone verifies that a new connection on a port with a
+// tombstoned (closed) conntrack entry is properly tracked. Without the fix,
+// updateIfExists treats tombstoned entries as live, causing track() to skip
+// creating a new connection. The subsequent SYN-ACK then fails IsValidInbound
+// because the entry is tombstoned, and the response packet gets dropped by ACL.
+func TestTCPPortReuseTombstone(t *testing.T) {
+	srcIP := netip.MustParseAddr("100.64.0.1")
+	dstIP := netip.MustParseAddr("100.64.0.2")
+	srcPort := uint16(12345)
+	dstPort := uint16(80)
+
+	t.Run("Outbound port reuse after graceful close", func(t *testing.T) {
+		tracker := NewTCPTracker(DefaultTCPTimeout, logger, flowLogger)
+		defer tracker.Close()
+
+		key := ConnKey{SrcIP: srcIP, DstIP: dstIP, SrcPort: srcPort, DstPort: dstPort}
+
+		// Establish and gracefully close a connection (server-initiated close)
+		establishConnection(t, tracker, srcIP, dstIP, srcPort, dstPort)
+
+		// Server sends FIN
+		valid := tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPFin|TCPAck, 0)
+		require.True(t, valid)
+
+		// Client sends FIN-ACK
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPFin|TCPAck, 0)
+
+		// Server sends final ACK
+		valid = tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPAck, 0)
+		require.True(t, valid)
+
+		// Connection should be tombstoned
+		conn := tracker.connections[key]
+		require.NotNil(t, conn, "old connection should still be in map")
+		require.True(t, conn.IsTombstone(), "old connection should be tombstoned")
+
+		// Now reuse the same port for a new connection
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPSyn, 100)
+
+		// The old tombstoned entry should be replaced with a new one
+		newConn := tracker.connections[key]
+		require.NotNil(t, newConn, "new connection should exist")
+		require.False(t, newConn.IsTombstone(), "new connection should not be tombstoned")
+		require.Equal(t, TCPStateSynSent, newConn.GetState())
+
+		// SYN-ACK for the new connection should be valid
+		valid = tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPSyn|TCPAck, 100)
+		require.True(t, valid, "SYN-ACK for new connection on reused port should be accepted")
+		require.Equal(t, TCPStateEstablished, newConn.GetState())
+
+		// Data transfer should work
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPAck, 100)
+		valid = tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPPush|TCPAck, 500)
+		require.True(t, valid, "data should be allowed on new connection")
+	})
+
+	t.Run("Outbound port reuse after RST", func(t *testing.T) {
+		tracker := NewTCPTracker(DefaultTCPTimeout, logger, flowLogger)
+		defer tracker.Close()
+
+		key := ConnKey{SrcIP: srcIP, DstIP: dstIP, SrcPort: srcPort, DstPort: dstPort}
+
+		// Establish and RST a connection
+		establishConnection(t, tracker, srcIP, dstIP, srcPort, dstPort)
+		valid := tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPRst|TCPAck, 0)
+		require.True(t, valid)
+
+		conn := tracker.connections[key]
+		require.True(t, conn.IsTombstone(), "RST connection should be tombstoned")
+
+		// Reuse the same port
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPSyn, 100)
+
+		newConn := tracker.connections[key]
+		require.NotNil(t, newConn)
+		require.False(t, newConn.IsTombstone())
+		require.Equal(t, TCPStateSynSent, newConn.GetState())
+
+		valid = tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPSyn|TCPAck, 100)
+		require.True(t, valid, "SYN-ACK should be accepted after RST tombstone")
+	})
+
+	t.Run("Inbound port reuse after close", func(t *testing.T) {
+		tracker := NewTCPTracker(DefaultTCPTimeout, logger, flowLogger)
+		defer tracker.Close()
+
+		clientIP := srcIP
+		serverIP := dstIP
+		clientPort := srcPort
+		serverPort := dstPort
+		key := ConnKey{SrcIP: clientIP, DstIP: serverIP, SrcPort: clientPort, DstPort: serverPort}
+
+		// Inbound connection: client SYN → server SYN-ACK → client ACK
+		tracker.TrackInbound(clientIP, serverIP, clientPort, serverPort, TCPSyn, nil, 100, 0)
+		tracker.TrackOutbound(serverIP, clientIP, serverPort, clientPort, TCPSyn|TCPAck, 100)
+		tracker.TrackInbound(clientIP, serverIP, clientPort, serverPort, TCPAck, nil, 100, 0)
+
+		conn := tracker.connections[key]
+		require.Equal(t, TCPStateEstablished, conn.GetState())
+
+		// Server-initiated close to reach Closed/tombstoned:
+		// Server FIN (opposite dir) → CloseWait
+		tracker.TrackOutbound(serverIP, clientIP, serverPort, clientPort, TCPFin|TCPAck, 100)
+		require.Equal(t, TCPStateCloseWait, conn.GetState())
+		// Client FIN-ACK (same dir as conn) → LastAck
+		tracker.TrackInbound(clientIP, serverIP, clientPort, serverPort, TCPFin|TCPAck, nil, 100, 0)
+		require.Equal(t, TCPStateLastAck, conn.GetState())
+		// Server final ACK (opposite dir) → Closed → tombstoned
+		tracker.TrackOutbound(serverIP, clientIP, serverPort, clientPort, TCPAck, 100)
+
+		require.True(t, conn.IsTombstone())
+
+		// New inbound connection on same ports
+		tracker.TrackInbound(clientIP, serverIP, clientPort, serverPort, TCPSyn, nil, 100, 0)
+
+		newConn := tracker.connections[key]
+		require.NotNil(t, newConn)
+		require.False(t, newConn.IsTombstone())
+		require.Equal(t, TCPStateSynReceived, newConn.GetState())
+
+		// Complete handshake: server SYN-ACK, then client ACK
+		tracker.TrackOutbound(serverIP, clientIP, serverPort, clientPort, TCPSyn|TCPAck, 100)
+		tracker.TrackInbound(clientIP, serverIP, clientPort, serverPort, TCPAck, nil, 100, 0)
+		require.Equal(t, TCPStateEstablished, newConn.GetState())
+	})
+
+	t.Run("Late ACK on tombstoned connection is harmless", func(t *testing.T) {
+		tracker := NewTCPTracker(DefaultTCPTimeout, logger, flowLogger)
+		defer tracker.Close()
+
+		key := ConnKey{SrcIP: srcIP, DstIP: dstIP, SrcPort: srcPort, DstPort: dstPort}
+
+		// Establish and close via passive close (server-initiated FIN → Closed → tombstoned)
+		establishConnection(t, tracker, srcIP, dstIP, srcPort, dstPort)
+		tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPFin|TCPAck, 0) // CloseWait
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPFin|TCPAck, 0)  // LastAck
+		tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPAck, 0)        // Closed
+
+		conn := tracker.connections[key]
+		require.True(t, conn.IsTombstone())
+
+		// Late ACK should be rejected (tombstoned)
+		valid := tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPAck, 0)
+		require.False(t, valid, "late ACK on tombstoned connection should be rejected")
+
+		// Late outbound ACK should not create a new connection (not a SYN)
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPAck, 0)
+		require.True(t, tracker.connections[key].IsTombstone(), "late outbound ACK should not replace tombstoned entry")
+	})
+}
+
+func TestTCPPortReuseTimeWait(t *testing.T) {
+	srcIP := netip.MustParseAddr("100.64.0.1")
+	dstIP := netip.MustParseAddr("100.64.0.2")
+	srcPort := uint16(12345)
+	dstPort := uint16(80)
+
+	t.Run("Outbound port reuse during TIME-WAIT (active close)", func(t *testing.T) {
+		tracker := NewTCPTracker(DefaultTCPTimeout, logger, flowLogger)
+		defer tracker.Close()
+
+		key := ConnKey{SrcIP: srcIP, DstIP: dstIP, SrcPort: srcPort, DstPort: dstPort}
+
+		// Establish connection
+		establishConnection(t, tracker, srcIP, dstIP, srcPort, dstPort)
+
+		// Active close: client (outbound initiator) sends FIN first
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPFin|TCPAck, 0)
+		conn := tracker.connections[key]
+		require.Equal(t, TCPStateFinWait1, conn.GetState())
+
+		// Server ACKs the FIN
+		valid := tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPAck, 0)
+		require.True(t, valid)
+		require.Equal(t, TCPStateFinWait2, conn.GetState())
+
+		// Server sends its own FIN
+		valid = tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPFin|TCPAck, 0)
+		require.True(t, valid)
+		require.Equal(t, TCPStateTimeWait, conn.GetState())
+
+		// Client sends final ACK (TIME-WAIT stays, not tombstoned)
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPAck, 0)
+		require.False(t, conn.IsTombstone(), "TIME-WAIT should not be tombstoned")
+
+		// New outbound SYN on the same port (port reuse during TIME-WAIT)
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPSyn, 100)
+
+		// Per RFC 1122/6191, new SYN during TIME-WAIT should start a new connection
+		newConn := tracker.connections[key]
+		require.NotNil(t, newConn, "new connection should exist")
+		require.False(t, newConn.IsTombstone(), "new connection should not be tombstoned")
+		require.Equal(t, TCPStateSynSent, newConn.GetState(), "new connection should be in SYN-SENT")
+
+		// SYN-ACK for new connection should be valid
+		valid = tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPSyn|TCPAck, 100)
+		require.True(t, valid, "SYN-ACK for new connection should be accepted")
+		require.Equal(t, TCPStateEstablished, newConn.GetState())
+	})
+
+	t.Run("Inbound SYN during TIME-WAIT falls through to normal tracking", func(t *testing.T) {
+		tracker := NewTCPTracker(DefaultTCPTimeout, logger, flowLogger)
+		defer tracker.Close()
+
+		key := ConnKey{SrcIP: srcIP, DstIP: dstIP, SrcPort: srcPort, DstPort: dstPort}
+
+		// Establish outbound connection and close via active close → TIME-WAIT
+		establishConnection(t, tracker, srcIP, dstIP, srcPort, dstPort)
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPFin|TCPAck, 0)
+		tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPAck, 0)
+		tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPFin|TCPAck, 0)
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPAck, 0)
+
+		conn := tracker.connections[key]
+		require.Equal(t, TCPStateTimeWait, conn.GetState())
+
+		// Inbound SYN on same ports during TIME-WAIT: IsValidInbound returns false
+		// so the filter falls through to ACL check + TrackInbound (which creates
+		// a new connection via track() → updateIfExists skips TIME-WAIT for SYN)
+		valid := tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPSyn, 0)
+		require.False(t, valid, "inbound SYN during TIME-WAIT should fail conntrack validation")
+
+		// Simulate what the filter does next: TrackInbound via the normal path
+		tracker.TrackInbound(dstIP, srcIP, dstPort, srcPort, TCPSyn, nil, 100, 0)
+
+		// The new inbound connection uses the inverted key (dst→src becomes src→dst in track)
+		invertedKey := ConnKey{SrcIP: dstIP, DstIP: srcIP, SrcPort: dstPort, DstPort: srcPort}
+		newConn := tracker.connections[invertedKey]
+		require.NotNil(t, newConn, "new inbound connection should be tracked")
+		require.Equal(t, TCPStateSynReceived, newConn.GetState())
+		require.False(t, newConn.IsTombstone())
+	})
+
+	t.Run("Late retransmit during TIME-WAIT still allowed", func(t *testing.T) {
+		tracker := NewTCPTracker(DefaultTCPTimeout, logger, flowLogger)
+		defer tracker.Close()
+
+		key := ConnKey{SrcIP: srcIP, DstIP: dstIP, SrcPort: srcPort, DstPort: dstPort}
+
+		// Establish and active close → TIME-WAIT
+		establishConnection(t, tracker, srcIP, dstIP, srcPort, dstPort)
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPFin|TCPAck, 0)
+		tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPAck, 0)
+		tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPFin|TCPAck, 0)
+		tracker.TrackOutbound(srcIP, dstIP, srcPort, dstPort, TCPAck, 0)
+
+		conn := tracker.connections[key]
+		require.Equal(t, TCPStateTimeWait, conn.GetState())
+
+		// Late ACK retransmits during TIME-WAIT should still be accepted
+		valid := tracker.IsValidInbound(dstIP, srcIP, dstPort, srcPort, TCPAck, 0)
+		require.True(t, valid, "retransmitted ACK during TIME-WAIT should be accepted")
+	})
+}
+
 func TestTCPTimeoutHandling(t *testing.T) {
 	// Create tracker with a very short timeout for testing
 	shortTimeout := 100 * time.Millisecond

client/firewall/uspfilter/filter.go

@@ -1,6 +1,7 @@
 package uspfilter
 
 import (
+	"context"
 	"encoding/binary"
 	"errors"
 	"fmt"
@@ -12,11 +13,13 @@ import (
 	"strings"
 	"sync"
 	"sync/atomic"
+	"time"
 
 	"github.com/google/gopacket"
 	"github.com/google/gopacket/layers"
 	"github.com/google/uuid"
 	log "github.com/sirupsen/logrus"
+	"golang.org/x/exp/maps"
 
 	firewall "github.com/netbirdio/netbird/client/firewall/manager"
 	"github.com/netbirdio/netbird/client/firewall/uspfilter/common"
@@ -24,15 +27,18 @@ import (
 	"github.com/netbirdio/netbird/client/firewall/uspfilter/forwarder"
 	nblog "github.com/netbirdio/netbird/client/firewall/uspfilter/log"
 	"github.com/netbirdio/netbird/client/iface/netstack"
+	nbid "github.com/netbirdio/netbird/client/internal/acl/id"
 	nftypes "github.com/netbirdio/netbird/client/internal/netflow/types"
 	"github.com/netbirdio/netbird/client/internal/statemanager"
 )
 
 const (
-	layerTypeAll = 0
+	layerTypeAll = 255
 
-	// ipTCPHeaderMinSize represents minimum IP (20) + TCP (20) header size for MSS calculation
-	ipTCPHeaderMinSize = 40
+	// ipv4TCPHeaderMinSize represents minimum IPv4 (20) + TCP (20) header size for MSS calculation
+	ipv4TCPHeaderMinSize = 40
+	// ipv6TCPHeaderMinSize represents minimum IPv6 (40) + TCP (20) header size for MSS calculation
+	ipv6TCPHeaderMinSize = 60
 )
 
 // serviceKey represents a protocol/port combination for netstack service registry
@@ -89,6 +95,7 @@ type Manager struct {
 	incomingDenyRules map[netip.Addr]RuleSet
 	incomingRules     map[netip.Addr]RuleSet
 	routeRules        RouteRules
+	routeRulesMap     map[nbid.RuleID]*RouteRule
 	decoders          sync.Pool
 	wgIface           common.IFaceMapper
 	nativeFirewall    firewall.Manager
@@ -132,9 +139,10 @@ type Manager struct {
 	netstackServices     map[serviceKey]struct{}
 	netstackServiceMutex sync.RWMutex
 
-	mtu             uint16
-	mssClampValue   uint16
-	mssClampEnabled bool
+	mtu               uint16
+	mssClampValueIPv4 uint16
+	mssClampValueIPv6 uint16
+	mssClampEnabled   bool
 }
 
 // decoder for packages
@@ -147,11 +155,28 @@ type decoder struct {
 	icmp4   layers.ICMPv4
 	icmp6   layers.ICMPv6
 	decoded []gopacket.LayerType
-	parser  *gopacket.DecodingLayerParser
+	parser4 *gopacket.DecodingLayerParser
+	parser6 *gopacket.DecodingLayerParser
 
 	dnatOrigPort uint16
 }
 
+// decodePacket decodes packet data using the appropriate parser based on IP version.
+func (d *decoder) decodePacket(data []byte) error {
+	if len(data) == 0 {
+		return errors.New("empty packet")
+	}
+	version := data[0] >> 4
+	switch version {
+	case 4:
+		return d.parser4.DecodeLayers(data, &d.decoded)
+	case 6:
+		return d.parser6.DecodeLayers(data, &d.decoded)
+	default:
+		return fmt.Errorf("unknown IP version %d", version)
+	}
+}
+
 // Create userspace firewall manager constructor
 func Create(iface common.IFaceMapper, disableServerRoutes bool, flowLogger nftypes.FlowLogger, mtu uint16) (*Manager, error) {
 	return create(iface, nil, disableServerRoutes, flowLogger, mtu)
@@ -209,11 +234,17 @@ func create(iface common.IFaceMapper, nativeFirewall firewall.Manager, disableSe
 				d := &decoder{
 					decoded: []gopacket.LayerType{},
 				}
-				d.parser = gopacket.NewDecodingLayerParser(
+				d.parser4 = gopacket.NewDecodingLayerParser(
 					layers.LayerTypeIPv4,
 					&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
 				)
-				d.parser.IgnoreUnsupported = true
+				d.parser4.IgnoreUnsupported = true
+
+				d.parser6 = gopacket.NewDecodingLayerParser(
+					layers.LayerTypeIPv6,
+					&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
+				)
+				d.parser6.IgnoreUnsupported = true
 				return d
 			},
 		},
@@ -229,6 +260,7 @@ func create(iface common.IFaceMapper, nativeFirewall firewall.Manager, disableSe
 		flowLogger:          flowLogger,
 		netstack:            netstack.IsEnabled(),
 		localForwarding:     enableLocalForwarding,
+		routeRulesMap:       make(map[nbid.RuleID]*RouteRule),
 		dnatMappings:        make(map[netip.Addr]netip.Addr),
 		portDNATRules:       []portDNATRule{},
 		netstackServices:    make(map[serviceKey]struct{}),
@@ -238,7 +270,8 @@ func create(iface common.IFaceMapper, nativeFirewall firewall.Manager, disableSe
 
 	if !disableMSSClamping {
 		m.mssClampEnabled = true
-		m.mssClampValue = mtu - ipTCPHeaderMinSize
+		m.mssClampValueIPv4 = mtu - ipv4TCPHeaderMinSize
+		m.mssClampValueIPv6 = mtu - ipv6TCPHeaderMinSize
 	}
 	if err := m.localipmanager.UpdateLocalIPs(iface); err != nil {
 		return nil, fmt.Errorf("update local IPs: %w", err)
@@ -262,15 +295,17 @@ func create(iface common.IFaceMapper, nativeFirewall firewall.Manager, disableSe
 }
 
 func (m *Manager) blockInvalidRouted(iface common.IFaceMapper) (firewall.Rule, error) {
-	wgPrefix, err := netip.ParsePrefix(iface.Address().Network.String())
-	if err != nil {
-		return nil, fmt.Errorf("parse wireguard network: %w", err)
-	}
+	wgPrefix := iface.Address().Network
 	log.Debugf("blocking invalid routed traffic for %s", wgPrefix)
 
+	sources := []netip.Prefix{netip.PrefixFrom(netip.IPv4Unspecified(), 0)}
+	if v6 := iface.Address().IPv6Net; v6.IsValid() {
+		sources = append(sources, netip.PrefixFrom(netip.IPv6Unspecified(), 0))
+	}
+
 	rule, err := m.addRouteFiltering(
 		nil,
-		[]netip.Prefix{netip.PrefixFrom(netip.IPv4Unspecified(), 0)},
+		sources,
 		firewall.Network{Prefix: wgPrefix},
 		firewall.ProtocolALL,
 		nil,
@@ -278,7 +313,22 @@ func (m *Manager) blockInvalidRouted(iface common.IFaceMapper) (firewall.Rule, e
 		firewall.ActionDrop,
 	)
 	if err != nil {
-		return nil, fmt.Errorf("block wg nte : %w", err)
+		return nil, fmt.Errorf("block wg v4 net: %w", err)
+	}
+
+	if v6Net := iface.Address().IPv6Net; v6Net.IsValid() {
+		log.Debugf("blocking invalid routed traffic for %s", v6Net)
+		if _, err := m.addRouteFiltering(
+			nil,
+			sources,
+			firewall.Network{Prefix: v6Net},
+			firewall.ProtocolALL,
+			nil,
+			nil,
+			firewall.ActionDrop,
+		); err != nil {
+			return nil, fmt.Errorf("block wg v6 net: %w", err)
+		}
 	}
 
 	// TODO: Block networks that we're a client of
@@ -439,19 +489,7 @@ func (m *Manager) AddPeerFiltering(
 	r.sPort = sPort
 	r.dPort = dPort
 
-	switch proto {
-	case firewall.ProtocolTCP:
-		r.protoLayer = layers.LayerTypeTCP
-	case firewall.ProtocolUDP:
-		r.protoLayer = layers.LayerTypeUDP
-	case firewall.ProtocolICMP:
-		r.protoLayer = layers.LayerTypeICMPv4
-		if r.ipLayer == layers.LayerTypeIPv6 {
-			r.protoLayer = layers.LayerTypeICMPv6
-		}
-	case firewall.ProtocolALL:
-		r.protoLayer = layerTypeAll
-	}
+	r.protoLayer = protoToLayer(proto, r.ipLayer)
 
 	m.mutex.Lock()
 	var targetMap map[netip.Addr]RuleSet
@@ -495,17 +533,22 @@ func (m *Manager) addRouteFiltering(
 		return m.nativeFirewall.AddRouteFiltering(id, sources, destination, proto, sPort, dPort, action)
 	}
 
-	ruleID := uuid.New().String()
+	ruleKey := nbid.GenerateRouteRuleKey(sources, destination, proto, sPort, dPort, action)
+
+	if existingRule, ok := m.routeRulesMap[ruleKey]; ok {
+		return existingRule, nil
+	}
+
 	rule := RouteRule{
 		// TODO: consolidate these IDs
-		id:      ruleID,
-		mgmtId:  id,
-		sources: sources,
-		dstSet:  destination.Set,
-		proto:   proto,
-		srcPort: sPort,
-		dstPort: dPort,
-		action:  action,
+		id:         string(ruleKey),
+		mgmtId:     id,
+		sources:    sources,
+		dstSet:     destination.Set,
+		protoLayer: protoToLayer(proto, ipLayerFromPrefix(destination.Prefix)),
+		srcPort:    sPort,
+		dstPort:    dPort,
+		action:     action,
 	}
 	if destination.IsPrefix() {
 		rule.destinations = []netip.Prefix{destination.Prefix}
@@ -513,6 +556,7 @@ func (m *Manager) addRouteFiltering(
 
 	m.routeRules = append(m.routeRules, &rule)
 	m.routeRules.Sort()
+	m.routeRulesMap[ruleKey] = &rule
 
 	return &rule, nil
 }
@@ -529,15 +573,20 @@ func (m *Manager) deleteRouteRule(rule firewall.Rule) error {
 		return m.nativeFirewall.DeleteRouteRule(rule)
 	}
 
-	ruleID := rule.ID()
+	ruleKey := nbid.RuleID(rule.ID())
+	if _, ok := m.routeRulesMap[ruleKey]; !ok {
+		return fmt.Errorf("route rule not found: %s", ruleKey)
+	}
+
 	idx := slices.IndexFunc(m.routeRules, func(r *RouteRule) bool {
-		return r.id == ruleID
+		return r.id == string(ruleKey)
 	})
 	if idx < 0 {
-		return fmt.Errorf("route rule not found: %s", ruleID)
+		return fmt.Errorf("route rule not found in slice: %s", ruleKey)
 	}
 
 	m.routeRules = slices.Delete(m.routeRules, idx, idx+1)
+	delete(m.routeRulesMap, ruleKey)
 	return nil
 }
 
@@ -584,6 +633,48 @@ func (m *Manager) SetLegacyManagement(isLegacy bool) error {
 // Flush doesn't need to be implemented for this manager
 func (m *Manager) Flush() error { return nil }
 
+// resetState clears all firewall rules and closes connection trackers.
+// Must be called with m.mutex held.
+func (m *Manager) resetState() {
+	maps.Clear(m.outgoingRules)
+	maps.Clear(m.incomingDenyRules)
+	maps.Clear(m.incomingRules)
+	maps.Clear(m.routeRulesMap)
+	m.routeRules = m.routeRules[:0]
+
+	if m.udpTracker != nil {
+		m.udpTracker.Close()
+	}
+
+	if m.icmpTracker != nil {
+		m.icmpTracker.Close()
+	}
+
+	if m.tcpTracker != nil {
+		m.tcpTracker.Close()
+	}
+
+	if fwder := m.forwarder.Load(); fwder != nil {
+		fwder.Stop()
+	}
+
+	if m.logger != nil {
+		ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+		defer cancel()
+		if err := m.logger.Stop(ctx); err != nil {
+			log.Errorf("failed to shutdown logger: %v", err)
+		}
+	}
+}
+
+// SetupEBPFProxyNoTrack creates notrack rules for eBPF proxy loopback traffic.
+func (m *Manager) SetupEBPFProxyNoTrack(proxyPort, wgPort uint16) error {
+	if m.nativeFirewall == nil {
+		return nil
+	}
+	return m.nativeFirewall.SetupEBPFProxyNoTrack(proxyPort, wgPort)
+}
+
 // UpdateSet updates the rule destinations associated with the given set
 // by merging the existing prefixes with the new ones, then deduplicating.
 func (m *Manager) UpdateSet(set firewall.Set, prefixes []netip.Prefix) error {
@@ -606,11 +697,7 @@ func (m *Manager) UpdateSet(set firewall.Set, prefixes []netip.Prefix) error {
 	}
 
 	destinations := matches[0].destinations
-	for _, prefix := range prefixes {
-		if prefix.Addr().Is4() {
-			destinations = append(destinations, prefix)
-		}
-	}
+	destinations = append(destinations, prefixes...)
 
 	slices.SortFunc(destinations, func(a, b netip.Prefix) int {
 		cmp := a.Addr().Compare(b.Addr())
@@ -649,7 +736,7 @@ func (m *Manager) filterOutbound(packetData []byte, size int) bool {
 	d := m.decoders.Get().(*decoder)
 	defer m.decoders.Put(d)
 
-	if err := d.parser.DecodeLayers(packetData, &d.decoded); err != nil {
+	if err := d.decodePacket(packetData); err != nil {
 		return false
 	}
 
@@ -730,12 +817,28 @@ func (m *Manager) clampTCPMSS(packetData []byte, d *decoder) bool {
 		return false
 	}
 
+	var mssClampValue uint16
+	var ipHeaderSize int
+	switch d.decoded[0] {
+	case layers.LayerTypeIPv4:
+		mssClampValue = m.mssClampValueIPv4
+		ipHeaderSize = int(d.ip4.IHL) * 4
+		if ipHeaderSize < 20 {
+			return false
+		}
+	case layers.LayerTypeIPv6:
+		mssClampValue = m.mssClampValueIPv6
+		ipHeaderSize = 40
+	default:
+		return false
+	}
+
 	mssOptionIndex := -1
 	var currentMSS uint16
 	for i, opt := range d.tcp.Options {
 		if opt.OptionType == layers.TCPOptionKindMSS && len(opt.OptionData) == 2 {
 			currentMSS = binary.BigEndian.Uint16(opt.OptionData)
-			if currentMSS > m.mssClampValue {
+			if currentMSS > mssClampValue {
 				mssOptionIndex = i
 				break
 			}
@@ -746,20 +849,15 @@ func (m *Manager) clampTCPMSS(packetData []byte, d *decoder) bool {
 		return false
 	}
 
-	ipHeaderSize := int(d.ip4.IHL) * 4
-	if ipHeaderSize < 20 {
+	if !m.updateMSSOption(packetData, d, mssOptionIndex, mssClampValue, ipHeaderSize) {
 		return false
 	}
 
-	if !m.updateMSSOption(packetData, d, mssOptionIndex, ipHeaderSize) {
-		return false
-	}
-
-	m.logger.Trace2("Clamped TCP MSS from %d to %d", currentMSS, m.mssClampValue)
+	m.logger.Trace2("Clamped TCP MSS from %d to %d", currentMSS, mssClampValue)
 	return true
 }
 
-func (m *Manager) updateMSSOption(packetData []byte, d *decoder, mssOptionIndex, ipHeaderSize int) bool {
+func (m *Manager) updateMSSOption(packetData []byte, d *decoder, mssOptionIndex int, mssClampValue uint16, ipHeaderSize int) bool {
 	tcpHeaderStart := ipHeaderSize
 	tcpOptionsStart := tcpHeaderStart + 20
 
@@ -774,7 +872,7 @@ func (m *Manager) updateMSSOption(packetData []byte, d *decoder, mssOptionIndex,
 	}
 
 	mssValueOffset := optOffset + 2
-	binary.BigEndian.PutUint16(packetData[mssValueOffset:mssValueOffset+2], m.mssClampValue)
+	binary.BigEndian.PutUint16(packetData[mssValueOffset:mssValueOffset+2], mssClampValue)
 
 	m.recalculateTCPChecksum(packetData, d, tcpHeaderStart)
 	return true
@@ -784,18 +882,32 @@ func (m *Manager) recalculateTCPChecksum(packetData []byte, d *decoder, tcpHeade
 	tcpLayer := packetData[tcpHeaderStart:]
 	tcpLength := len(packetData) - tcpHeaderStart
 
+	// Zero out existing checksum
 	tcpLayer[16] = 0
 	tcpLayer[17] = 0
 
+	// Build pseudo-header checksum based on IP version
 	var pseudoSum uint32
-	pseudoSum += uint32(d.ip4.SrcIP[0])<<8 | uint32(d.ip4.SrcIP[1])
-	pseudoSum += uint32(d.ip4.SrcIP[2])<<8 | uint32(d.ip4.SrcIP[3])
-	pseudoSum += uint32(d.ip4.DstIP[0])<<8 | uint32(d.ip4.DstIP[1])
-	pseudoSum += uint32(d.ip4.DstIP[2])<<8 | uint32(d.ip4.DstIP[3])
-	pseudoSum += uint32(d.ip4.Protocol)
-	pseudoSum += uint32(tcpLength)
+	switch d.decoded[0] {
+	case layers.LayerTypeIPv4:
+		pseudoSum += uint32(d.ip4.SrcIP[0])<<8 | uint32(d.ip4.SrcIP[1])
+		pseudoSum += uint32(d.ip4.SrcIP[2])<<8 | uint32(d.ip4.SrcIP[3])
+		pseudoSum += uint32(d.ip4.DstIP[0])<<8 | uint32(d.ip4.DstIP[1])
+		pseudoSum += uint32(d.ip4.DstIP[2])<<8 | uint32(d.ip4.DstIP[3])
+		pseudoSum += uint32(d.ip4.Protocol)
+		pseudoSum += uint32(tcpLength)
+	case layers.LayerTypeIPv6:
+		for i := 0; i < 16; i += 2 {
+			pseudoSum += uint32(d.ip6.SrcIP[i])<<8 | uint32(d.ip6.SrcIP[i+1])
+		}
+		for i := 0; i < 16; i += 2 {
+			pseudoSum += uint32(d.ip6.DstIP[i])<<8 | uint32(d.ip6.DstIP[i+1])
+		}
+		pseudoSum += uint32(tcpLength)
+		pseudoSum += uint32(layers.IPProtocolTCP)
+	}
 
-	var sum uint32 = pseudoSum
+	sum := pseudoSum
 	for i := 0; i < tcpLength-1; i += 2 {
 		sum += uint32(tcpLayer[i])<<8 | uint32(tcpLayer[i+1])
 	}
@@ -833,6 +945,9 @@ func (m *Manager) trackOutbound(d *decoder, srcIP, dstIP netip.Addr, packetData
 		}
 	case layers.LayerTypeICMPv4:
 		m.icmpTracker.TrackOutbound(srcIP, dstIP, d.icmp4.Id, d.icmp4.TypeCode, d.icmp4.Payload, size)
+	case layers.LayerTypeICMPv6:
+		id, tc := icmpv6EchoFields(d)
+		m.icmpTracker.TrackOutbound(srcIP, dstIP, id, tc, d.icmp6.Payload, size)
 	}
 }
 
@@ -846,6 +961,9 @@ func (m *Manager) trackInbound(d *decoder, srcIP, dstIP netip.Addr, ruleID []byt
 		m.tcpTracker.TrackInbound(srcIP, dstIP, uint16(d.tcp.SrcPort), uint16(d.tcp.DstPort), flags, ruleID, size, d.dnatOrigPort)
 	case layers.LayerTypeICMPv4:
 		m.icmpTracker.TrackInbound(srcIP, dstIP, d.icmp4.Id, d.icmp4.TypeCode, ruleID, d.icmp4.Payload, size)
+	case layers.LayerTypeICMPv6:
+		id, tc := icmpv6EchoFields(d)
+		m.icmpTracker.TrackInbound(srcIP, dstIP, id, tc, ruleID, d.icmp6.Payload, size)
 	}
 
 	d.dnatOrigPort = 0
@@ -905,15 +1023,19 @@ func (m *Manager) filterInbound(packetData []byte, size int) bool {
 
 	// TODO: pass fragments of routed packets to forwarder
 	if fragment {
-		m.logger.Trace4("packet is a fragment: src=%v dst=%v id=%v flags=%v",
-			srcIP, dstIP, d.ip4.Id, d.ip4.Flags)
+		if d.decoded[0] == layers.LayerTypeIPv4 {
+			m.logger.Trace4("packet is a fragment: src=%v dst=%v id=%v flags=%v",
+				srcIP, dstIP, d.ip4.Id, d.ip4.Flags)
+		} else {
+			m.logger.Trace2("packet is an IPv6 fragment: src=%v dst=%v", srcIP, dstIP)
+		}
 		return false
 	}
 
 	// TODO: optimize port DNAT by caching matched rules in conntrack
 	if translated := m.translateInboundPortDNAT(packetData, d, srcIP, dstIP); translated {
 		// Re-decode after port DNAT translation to update port information
-		if err := d.parser.DecodeLayers(packetData, &d.decoded); err != nil {
+		if err := d.decodePacket(packetData); err != nil {
 			m.logger.Error1("failed to re-decode packet after port DNAT: %v", err)
 			return true
 		}
@@ -922,7 +1044,7 @@ func (m *Manager) filterInbound(packetData []byte, size int) bool {
 
 	if translated := m.translateInboundReverse(packetData, d); translated {
 		// Re-decode after translation to get original addresses
-		if err := d.parser.DecodeLayers(packetData, &d.decoded); err != nil {
+		if err := d.decodePacket(packetData); err != nil {
 			m.logger.Error1("failed to re-decode packet after reverse DNAT: %v", err)
 			return true
 		}
@@ -945,7 +1067,7 @@ func (m *Manager) filterInbound(packetData []byte, size int) bool {
 func (m *Manager) handleLocalTraffic(d *decoder, srcIP, dstIP netip.Addr, packetData []byte, size int) bool {
 	ruleID, blocked := m.peerACLsBlock(srcIP, d, packetData)
 	if blocked {
-		_, pnum := getProtocolFromPacket(d)
+		pnum := getProtocolFromPacket(d)
 		srcPort, dstPort := getPortsFromPacket(d)
 
 		m.logger.Trace6("Dropping local packet (ACL denied): rule_id=%s proto=%v src=%s:%d dst=%s:%d",
@@ -1010,20 +1132,22 @@ func (m *Manager) handleRoutedTraffic(d *decoder, srcIP, dstIP netip.Addr, packe
 		return false
 	}
 
-	proto, pnum := getProtocolFromPacket(d)
+	protoLayer := d.decoded[1]
 	srcPort, dstPort := getPortsFromPacket(d)
 
-	ruleID, pass := m.routeACLsPass(srcIP, dstIP, proto, srcPort, dstPort)
+	ruleID, pass := m.routeACLsPass(srcIP, dstIP, protoLayer, srcPort, dstPort)
 	if !pass {
+		proto := getProtocolFromPacket(d)
+
 		m.logger.Trace6("Dropping routed packet (ACL denied): rule_id=%s proto=%v src=%s:%d dst=%s:%d",
-			ruleID, pnum, srcIP, srcPort, dstIP, dstPort)
+			ruleID, proto, srcIP, srcPort, dstIP, dstPort)
 
 		m.flowLogger.StoreEvent(nftypes.EventFields{
 			FlowID:     uuid.New(),
 			Type:       nftypes.TypeDrop,
 			RuleID:     ruleID,
 			Direction:  nftypes.Ingress,
-			Protocol:   pnum,
+			Protocol:   proto,
 			SourceIP:   srcIP,
 			DestIP:     dstIP,
 			SourcePort: srcPort,
@@ -1052,16 +1176,77 @@ func (m *Manager) handleRoutedTraffic(d *decoder, srcIP, dstIP netip.Addr, packe
 	return true
 }
 
-func getProtocolFromPacket(d *decoder) (firewall.Protocol, nftypes.Protocol) {
+// icmpv6EchoFields extracts the echo identifier from an ICMPv6 packet and maps
+// the ICMPv6 type code to an ICMPv4TypeCode so the ICMP conntrack can handle
+// both families uniformly. The echo ID is in the first two payload bytes.
+func icmpv6EchoFields(d *decoder) (id uint16, tc layers.ICMPv4TypeCode) {
+	if len(d.icmp6.Payload) >= 2 {
+		id = uint16(d.icmp6.Payload[0])<<8 | uint16(d.icmp6.Payload[1])
+	}
+	// Map ICMPv6 echo types to ICMPv4 equivalents for unified tracking.
+	switch d.icmp6.TypeCode.Type() {
+	case layers.ICMPv6TypeEchoRequest:
+		tc = layers.CreateICMPv4TypeCode(layers.ICMPv4TypeEchoRequest, 0)
+	case layers.ICMPv6TypeEchoReply:
+		tc = layers.CreateICMPv4TypeCode(layers.ICMPv4TypeEchoReply, 0)
+	default:
+		tc = layers.CreateICMPv4TypeCode(d.icmp6.TypeCode.Type(), d.icmp6.TypeCode.Code())
+	}
+	return id, tc
+}
+
+// protoLayerMatches checks if a packet's protocol layer matches a rule's expected
+// protocol layer. ICMPv4 and ICMPv6 are treated as equivalent when matching
+// ICMP rules since management sends a single ICMP rule for both families.
+func protoLayerMatches(ruleLayer, packetLayer gopacket.LayerType) bool {
+	if ruleLayer == packetLayer {
+		return true
+	}
+	if ruleLayer == layers.LayerTypeICMPv4 && packetLayer == layers.LayerTypeICMPv6 {
+		return true
+	}
+	if ruleLayer == layers.LayerTypeICMPv6 && packetLayer == layers.LayerTypeICMPv4 {
+		return true
+	}
+	return false
+}
+
+func ipLayerFromPrefix(p netip.Prefix) gopacket.LayerType {
+	if p.Addr().Is6() {
+		return layers.LayerTypeIPv6
+	}
+	return layers.LayerTypeIPv4
+}
+
+func protoToLayer(proto firewall.Protocol, ipLayer gopacket.LayerType) gopacket.LayerType {
+	switch proto {
+	case firewall.ProtocolTCP:
+		return layers.LayerTypeTCP
+	case firewall.ProtocolUDP:
+		return layers.LayerTypeUDP
+	case firewall.ProtocolICMP:
+		if ipLayer == layers.LayerTypeIPv6 {
+			return layers.LayerTypeICMPv6
+		}
+		return layers.LayerTypeICMPv4
+	case firewall.ProtocolALL:
+		return layerTypeAll
+	}
+	return 0
+}
+
+func getProtocolFromPacket(d *decoder) nftypes.Protocol {
 	switch d.decoded[1] {
 	case layers.LayerTypeTCP:
-		return firewall.ProtocolTCP, nftypes.TCP
+		return nftypes.TCP
 	case layers.LayerTypeUDP:
-		return firewall.ProtocolUDP, nftypes.UDP
-	case layers.LayerTypeICMPv4, layers.LayerTypeICMPv6:
-		return firewall.ProtocolICMP, nftypes.ICMP
+		return nftypes.UDP
+	case layers.LayerTypeICMPv4:
+		return nftypes.ICMP
+	case layers.LayerTypeICMPv6:
+		return nftypes.ICMPv6
 	default:
-		return firewall.ProtocolALL, nftypes.ProtocolUnknown
+		return nftypes.ProtocolUnknown
 	}
 }
 
@@ -1080,7 +1265,7 @@ func getPortsFromPacket(d *decoder) (srcPort, dstPort uint16) {
 // It returns true, false if the packet is valid and not a fragment.
 // It returns true, true if the packet is a fragment and valid.
 func (m *Manager) isValidPacket(d *decoder, packetData []byte) (bool, bool) {
-	if err := d.parser.DecodeLayers(packetData, &d.decoded); err != nil {
+	if err := d.decodePacket(packetData); err != nil {
 		m.logger.Trace1("couldn't decode packet, err: %s", err)
 		return false, false
 	}
@@ -1093,10 +1278,18 @@ func (m *Manager) isValidPacket(d *decoder, packetData []byte) (bool, bool) {
 	}
 
 	// Fragments are also valid
-	if l == 1 && d.decoded[0] == layers.LayerTypeIPv4 {
-		ip4 := d.ip4
-		if ip4.Flags&layers.IPv4MoreFragments != 0 || ip4.FragOffset != 0 {
-			return true, true
+	if l == 1 {
+		switch d.decoded[0] {
+		case layers.LayerTypeIPv4:
+			if d.ip4.Flags&layers.IPv4MoreFragments != 0 || d.ip4.FragOffset != 0 {
+				return true, true
+			}
+		case layers.LayerTypeIPv6:
+			// IPv6 uses Fragment extension header (NextHeader=44). If gopacket
+			// only decoded the IPv6 layer, the transport is in a fragment.
+			if d.ip6.NextHeader == layers.IPProtocolIPv6Fragment {
+				return true, true
+			}
 		}
 	}
 
@@ -1134,21 +1327,34 @@ func (m *Manager) isValidTrackedConnection(d *decoder, srcIP, dstIP netip.Addr,
 			size,
 		)
 
-		// TODO: ICMPv6
+	case layers.LayerTypeICMPv6:
+		id, _ := icmpv6EchoFields(d)
+		return m.icmpTracker.IsValidInbound(
+			srcIP,
+			dstIP,
+			id,
+			d.icmp6.TypeCode.Type(),
+			size,
+		)
 	}
 
 	return false
 }
 
-// isSpecialICMP returns true if the packet is a special ICMP packet that should be allowed
+// isSpecialICMP returns true if the packet is a special ICMP error packet that should be allowed.
 func (m *Manager) isSpecialICMP(d *decoder) bool {
-	if d.decoded[1] != layers.LayerTypeICMPv4 {
-		return false
+	switch d.decoded[1] {
+	case layers.LayerTypeICMPv4:
+		icmpType := d.icmp4.TypeCode.Type()
+		return icmpType == layers.ICMPv4TypeDestinationUnreachable ||
+			icmpType == layers.ICMPv4TypeTimeExceeded
+	case layers.LayerTypeICMPv6:
+		icmpType := d.icmp6.TypeCode.Type()
+		return icmpType == layers.ICMPv6TypeDestinationUnreachable ||
+			icmpType == layers.ICMPv6TypePacketTooBig ||
+			icmpType == layers.ICMPv6TypeTimeExceeded
 	}
-
-	icmpType := d.icmp4.TypeCode.Type()
-	return icmpType == layers.ICMPv4TypeDestinationUnreachable ||
-		icmpType == layers.ICMPv4TypeTimeExceeded
+	return false
 }
 
 func (m *Manager) peerACLsBlock(srcIP netip.Addr, d *decoder, packetData []byte) ([]byte, bool) {
@@ -1205,7 +1411,7 @@ func validateRule(ip netip.Addr, packetData []byte, rules map[string]PeerRule, d
 			return rule.mgmtId, rule.drop, true
 		}
 
-		if payloadLayer != rule.protoLayer {
+		if !protoLayerMatches(rule.protoLayer, payloadLayer) {
 			continue
 		}
 
@@ -1233,19 +1439,29 @@ func validateRule(ip netip.Addr, packetData []byte, rules map[string]PeerRule, d
 }
 
 // routeACLsPass returns true if the packet is allowed by the route ACLs
-func (m *Manager) routeACLsPass(srcIP, dstIP netip.Addr, proto firewall.Protocol, srcPort, dstPort uint16) ([]byte, bool) {
+func (m *Manager) routeACLsPass(srcIP, dstIP netip.Addr, protoLayer gopacket.LayerType, srcPort, dstPort uint16) ([]byte, bool) {
 	m.mutex.RLock()
 	defer m.mutex.RUnlock()
 
 	for _, rule := range m.routeRules {
-		if matches := m.ruleMatches(rule, srcIP, dstIP, proto, srcPort, dstPort); matches {
+		if matches := m.ruleMatches(rule, srcIP, dstIP, protoLayer, srcPort, dstPort); matches {
 			return rule.mgmtId, rule.action == firewall.ActionAccept
 		}
 	}
 	return nil, false
 }
 
-func (m *Manager) ruleMatches(rule *RouteRule, srcAddr, dstAddr netip.Addr, proto firewall.Protocol, srcPort, dstPort uint16) bool {
+func (m *Manager) ruleMatches(rule *RouteRule, srcAddr, dstAddr netip.Addr, protoLayer gopacket.LayerType, srcPort, dstPort uint16) bool {
+	if rule.protoLayer != layerTypeAll && !protoLayerMatches(rule.protoLayer, protoLayer) {
+		return false
+	}
+
+	if protoLayer == layers.LayerTypeTCP || protoLayer == layers.LayerTypeUDP {
+		if !portsMatch(rule.srcPort, srcPort) || !portsMatch(rule.dstPort, dstPort) {
+			return false
+		}
+	}
+
 	destMatched := false
 	for _, dst := range rule.destinations {
 		if dst.Contains(dstAddr) {
@@ -1264,21 +1480,8 @@ func (m *Manager) ruleMatches(rule *RouteRule, srcAddr, dstAddr netip.Addr, prot
 			break
 		}
 	}
-	if !sourceMatched {
-		return false
-	}
 
-	if rule.proto != firewall.ProtocolALL && rule.proto != proto {
-		return false
-	}
-
-	if proto == firewall.ProtocolTCP || proto == firewall.ProtocolUDP {
-		if !portsMatch(rule.srcPort, srcPort) || !portsMatch(rule.dstPort, dstPort) {
-			return false
-		}
-	}
-
-	return true
+	return sourceMatched
 }
 
 // AddUDPPacketHook calls hook when UDP packet from given direction matched
@@ -1454,7 +1657,8 @@ func (m *Manager) shouldForward(d *decoder, dstIP netip.Addr) bool {
 	}
 
 	// traffic to our other local interfaces (not NetBird IP) - always forward
-	if dstIP != m.wgIface.Address().IP {
+	addr := m.wgIface.Address()
+	if dstIP != addr.IP && (!addr.IPv6.IsValid() || dstIP != addr.IPv6) {
 		return true
 	}
 

client/firewall/uspfilter/filter_bench_test.go

@@ -955,7 +955,7 @@ func BenchmarkRouteACLs(b *testing.B) {
 		for _, tc := range cases {
 			srcIP := netip.MustParseAddr(tc.srcIP)
 			dstIP := netip.MustParseAddr(tc.dstIP)
-			manager.routeACLsPass(srcIP, dstIP, tc.proto, 0, tc.dstPort)
+			manager.routeACLsPass(srcIP, dstIP, protoToLayer(tc.proto, layers.LayerTypeIPv4), 0, tc.dstPort)
 		}
 	}
 }
@@ -1023,7 +1023,8 @@ func BenchmarkMSSClamping(b *testing.B) {
 			}()
 
 			manager.mssClampEnabled = true
-			manager.mssClampValue = 1240
+			manager.mssClampValueIPv4 = 1240
+			manager.mssClampValueIPv6 = 1220
 
 			srcIP := net.ParseIP("100.64.0.2")
 			dstIP := net.ParseIP("8.8.8.8")
@@ -1088,7 +1089,8 @@ func BenchmarkMSSClampingOverhead(b *testing.B) {
 
 			manager.mssClampEnabled = sc.enabled
 			if sc.enabled {
-				manager.mssClampValue = 1240
+				manager.mssClampValueIPv4 = 1240
+				manager.mssClampValueIPv6 = 1220
 			}
 
 			srcIP := net.ParseIP("100.64.0.2")
@@ -1141,7 +1143,8 @@ func BenchmarkMSSClampingMemory(b *testing.B) {
 			}()
 
 			manager.mssClampEnabled = true
-			manager.mssClampValue = 1240
+			manager.mssClampValueIPv4 = 1240
+			manager.mssClampValueIPv6 = 1220
 
 			srcIP := net.ParseIP("100.64.0.2")
 			dstIP := net.ParseIP("8.8.8.8")

client/firewall/uspfilter/filter_filter_test.go

@@ -539,53 +539,236 @@ func TestPeerACLFiltering(t *testing.T) {
 	}
 }
 
+func TestPeerACLFilteringIPv6(t *testing.T) {
+	localIP := netip.MustParseAddr("100.10.0.100")
+	localIPv6 := netip.MustParseAddr("fd00::100")
+	wgNet := netip.MustParsePrefix("100.10.0.0/16")
+	wgNetV6 := netip.MustParsePrefix("fd00::/64")
+
+	ifaceMock := &IFaceMock{
+		SetFilterFunc: func(device.PacketFilter) error { return nil },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      localIP,
+				Network: wgNet,
+				IPv6:    localIPv6,
+				IPv6Net: wgNetV6,
+			}
+		},
+	}
+
+	manager, err := Create(ifaceMock, false, flowLogger, iface.DefaultMTU)
+	require.NoError(t, err)
+	t.Cleanup(func() { require.NoError(t, manager.Close(nil)) })
+
+	err = manager.UpdateLocalIPs()
+	require.NoError(t, err)
+
+	testCases := []struct {
+		name            string
+		srcIP           string
+		dstIP           string
+		proto           fw.Protocol
+		srcPort         uint16
+		dstPort         uint16
+		ruleIP          string
+		ruleProto       fw.Protocol
+		ruleDstPort     *fw.Port
+		ruleAction      fw.Action
+		shouldBeBlocked bool
+	}{
+		{
+			name:            "IPv6: allow TCP from peer",
+			srcIP:           "fd00::1",
+			dstIP:           "fd00::100",
+			proto:           fw.ProtocolTCP,
+			srcPort:         12345,
+			dstPort:         443,
+			ruleIP:          "fd00::1",
+			ruleProto:       fw.ProtocolTCP,
+			ruleDstPort:     &fw.Port{Values: []uint16{443}},
+			ruleAction:      fw.ActionAccept,
+			shouldBeBlocked: false,
+		},
+		{
+			name:            "IPv6: allow UDP from peer",
+			srcIP:           "fd00::1",
+			dstIP:           "fd00::100",
+			proto:           fw.ProtocolUDP,
+			srcPort:         12345,
+			dstPort:         53,
+			ruleIP:          "fd00::1",
+			ruleProto:       fw.ProtocolUDP,
+			ruleDstPort:     &fw.Port{Values: []uint16{53}},
+			ruleAction:      fw.ActionAccept,
+			shouldBeBlocked: false,
+		},
+		{
+			name:            "IPv6: allow ICMPv6 from peer",
+			srcIP:           "fd00::1",
+			dstIP:           "fd00::100",
+			proto:           fw.ProtocolICMP,
+			ruleIP:          "fd00::1",
+			ruleProto:       fw.ProtocolICMP,
+			ruleAction:      fw.ActionAccept,
+			shouldBeBlocked: false,
+		},
+		{
+			name:            "IPv6: block TCP without rule",
+			srcIP:           "fd00::2",
+			dstIP:           "fd00::100",
+			proto:           fw.ProtocolTCP,
+			srcPort:         12345,
+			dstPort:         443,
+			ruleIP:          "fd00::1",
+			ruleProto:       fw.ProtocolTCP,
+			ruleDstPort:     &fw.Port{Values: []uint16{443}},
+			ruleAction:      fw.ActionAccept,
+			shouldBeBlocked: true,
+		},
+		{
+			name:            "IPv6: drop rule",
+			srcIP:           "fd00::1",
+			dstIP:           "fd00::100",
+			proto:           fw.ProtocolTCP,
+			srcPort:         12345,
+			dstPort:         22,
+			ruleIP:          "fd00::1",
+			ruleProto:       fw.ProtocolTCP,
+			ruleDstPort:     &fw.Port{Values: []uint16{22}},
+			ruleAction:      fw.ActionDrop,
+			shouldBeBlocked: true,
+		},
+		{
+			name:            "IPv6: allow all protocols",
+			srcIP:           "fd00::1",
+			dstIP:           "fd00::100",
+			proto:           fw.ProtocolUDP,
+			srcPort:         12345,
+			dstPort:         9999,
+			ruleIP:          "fd00::1",
+			ruleProto:       fw.ProtocolALL,
+			ruleAction:      fw.ActionAccept,
+			shouldBeBlocked: false,
+		},
+		{
+			name:            "IPv6: v4 wildcard ICMP rule matches ICMPv6 via protoLayerMatches",
+			srcIP:           "fd00::1",
+			dstIP:           "fd00::100",
+			proto:           fw.ProtocolICMP,
+			ruleIP:          "0.0.0.0",
+			ruleProto:       fw.ProtocolICMP,
+			ruleAction:      fw.ActionAccept,
+			shouldBeBlocked: false,
+		},
+	}
+
+	t.Run("IPv6 implicit DROP (no rules)", func(t *testing.T) {
+		packet := createTestPacket(t, "fd00::1", "fd00::100", fw.ProtocolTCP, 12345, 443)
+		isDropped := manager.FilterInbound(packet, 0)
+		require.True(t, isDropped, "IPv6 packet should be dropped when no rules exist")
+	})
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			if tc.ruleAction == fw.ActionDrop {
+				rules, err := manager.AddPeerFiltering(nil, net.ParseIP(tc.ruleIP), fw.ProtocolALL, nil, nil, fw.ActionAccept, "")
+				require.NoError(t, err)
+				t.Cleanup(func() {
+					for _, rule := range rules {
+						require.NoError(t, manager.DeletePeerRule(rule))
+					}
+				})
+			}
+
+			rules, err := manager.AddPeerFiltering(nil, net.ParseIP(tc.ruleIP), tc.ruleProto, nil, tc.ruleDstPort, tc.ruleAction, "")
+			require.NoError(t, err)
+			require.NotEmpty(t, rules)
+			t.Cleanup(func() {
+				for _, rule := range rules {
+					require.NoError(t, manager.DeletePeerRule(rule))
+				}
+			})
+
+			packet := createTestPacket(t, tc.srcIP, tc.dstIP, tc.proto, tc.srcPort, tc.dstPort)
+			isDropped := manager.FilterInbound(packet, 0)
+			require.Equal(t, tc.shouldBeBlocked, isDropped, "packet filter result mismatch")
+		})
+	}
+}
+
 func createTestPacket(t *testing.T, srcIP, dstIP string, proto fw.Protocol, srcPort, dstPort uint16) []byte {
 	t.Helper()
 
+	src := net.ParseIP(srcIP)
+	dst := net.ParseIP(dstIP)
+
 	buf := gopacket.NewSerializeBuffer()
 	opts := gopacket.SerializeOptions{
 		ComputeChecksums: true,
 		FixLengths:       true,
 	}
 
-	ipLayer := &layers.IPv4{
-		Version: 4,
-		TTL:     64,
-		SrcIP:   net.ParseIP(srcIP),
-		DstIP:   net.ParseIP(dstIP),
-	}
+	// Detect address family
+	isV6 := src.To4() == nil
 
 	var err error
-	switch proto {
-	case fw.ProtocolTCP:
-		ipLayer.Protocol = layers.IPProtocolTCP
-		tcp := &layers.TCP{
-			SrcPort: layers.TCPPort(srcPort),
-			DstPort: layers.TCPPort(dstPort),
-		}
-		err = tcp.SetNetworkLayerForChecksum(ipLayer)
-		require.NoError(t, err)
-		err = gopacket.SerializeLayers(buf, opts, ipLayer, tcp)
 
-	case fw.ProtocolUDP:
-		ipLayer.Protocol = layers.IPProtocolUDP
-		udp := &layers.UDP{
-			SrcPort: layers.UDPPort(srcPort),
-			DstPort: layers.UDPPort(dstPort),
+	if isV6 {
+		ip6 := &layers.IPv6{
+			Version:  6,
+			HopLimit: 64,
+			SrcIP:    src,
+			DstIP:    dst,
 		}
-		err = udp.SetNetworkLayerForChecksum(ipLayer)
-		require.NoError(t, err)
-		err = gopacket.SerializeLayers(buf, opts, ipLayer, udp)
 
-	case fw.ProtocolICMP:
-		ipLayer.Protocol = layers.IPProtocolICMPv4
-		icmp := &layers.ICMPv4{
-			TypeCode: layers.CreateICMPv4TypeCode(layers.ICMPv4TypeEchoRequest, 0),
+		switch proto {
+		case fw.ProtocolTCP:
+			ip6.NextHeader = layers.IPProtocolTCP
+			tcp := &layers.TCP{SrcPort: layers.TCPPort(srcPort), DstPort: layers.TCPPort(dstPort)}
+			_ = tcp.SetNetworkLayerForChecksum(ip6)
+			err = gopacket.SerializeLayers(buf, opts, ip6, tcp)
+		case fw.ProtocolUDP:
+			ip6.NextHeader = layers.IPProtocolUDP
+			udp := &layers.UDP{SrcPort: layers.UDPPort(srcPort), DstPort: layers.UDPPort(dstPort)}
+			_ = udp.SetNetworkLayerForChecksum(ip6)
+			err = gopacket.SerializeLayers(buf, opts, ip6, udp)
+		case fw.ProtocolICMP:
+			ip6.NextHeader = layers.IPProtocolICMPv6
+			icmp := &layers.ICMPv6{
+				TypeCode: layers.CreateICMPv6TypeCode(layers.ICMPv6TypeEchoRequest, 0),
+			}
+			_ = icmp.SetNetworkLayerForChecksum(ip6)
+			err = gopacket.SerializeLayers(buf, opts, ip6, icmp)
+		default:
+			err = gopacket.SerializeLayers(buf, opts, ip6)
+		}
+	} else {
+		ip4 := &layers.IPv4{
+			Version: 4,
+			TTL:     64,
+			SrcIP:   src,
+			DstIP:   dst,
 		}
-		err = gopacket.SerializeLayers(buf, opts, ipLayer, icmp)
 
-	default:
-		err = gopacket.SerializeLayers(buf, opts, ipLayer)
+		switch proto {
+		case fw.ProtocolTCP:
+			ip4.Protocol = layers.IPProtocolTCP
+			tcp := &layers.TCP{SrcPort: layers.TCPPort(srcPort), DstPort: layers.TCPPort(dstPort)}
+			_ = tcp.SetNetworkLayerForChecksum(ip4)
+			err = gopacket.SerializeLayers(buf, opts, ip4, tcp)
+		case fw.ProtocolUDP:
+			ip4.Protocol = layers.IPProtocolUDP
+			udp := &layers.UDP{SrcPort: layers.UDPPort(srcPort), DstPort: layers.UDPPort(dstPort)}
+			_ = udp.SetNetworkLayerForChecksum(ip4)
+			err = gopacket.SerializeLayers(buf, opts, ip4, udp)
+		case fw.ProtocolICMP:
+			ip4.Protocol = layers.IPProtocolICMPv4
+			icmp := &layers.ICMPv4{TypeCode: layers.CreateICMPv4TypeCode(layers.ICMPv4TypeEchoRequest, 0)}
+			err = gopacket.SerializeLayers(buf, opts, ip4, icmp)
+		default:
+			err = gopacket.SerializeLayers(buf, opts, ip4)
+		}
 	}
 
 	require.NoError(t, err)
@@ -1259,7 +1442,7 @@ func TestRouteACLFiltering(t *testing.T) {
 
 			// testing routeACLsPass only and not FilterInbound, as routed packets are dropped after being passed
 			// to the forwarder
-			_, isAllowed := manager.routeACLsPass(srcIP, dstIP, tc.proto, tc.srcPort, tc.dstPort)
+			_, isAllowed := manager.routeACLsPass(srcIP, dstIP, protoToLayer(tc.proto, layers.LayerTypeIPv4), tc.srcPort, tc.dstPort)
 			require.Equal(t, tc.shouldPass, isAllowed)
 		})
 	}
@@ -1445,7 +1628,7 @@ func TestRouteACLOrder(t *testing.T) {
 				srcIP := netip.MustParseAddr(p.srcIP)
 				dstIP := netip.MustParseAddr(p.dstIP)
 
-				_, isAllowed := manager.routeACLsPass(srcIP, dstIP, p.proto, p.srcPort, p.dstPort)
+				_, isAllowed := manager.routeACLsPass(srcIP, dstIP, protoToLayer(p.proto, layers.LayerTypeIPv4), p.srcPort, p.dstPort)
 				require.Equal(t, p.shouldPass, isAllowed, "packet %d failed", i)
 			}
 		})
@@ -1488,13 +1671,113 @@ func TestRouteACLSet(t *testing.T) {
 	dstIP := netip.MustParseAddr("192.168.1.100")
 
 	// Check that traffic is dropped (empty set shouldn't match anything)
-	_, isAllowed := manager.routeACLsPass(srcIP, dstIP, fw.ProtocolTCP, 12345, 80)
+	_, isAllowed := manager.routeACLsPass(srcIP, dstIP, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
 	require.False(t, isAllowed, "Empty set should not allow any traffic")
 
 	err = manager.UpdateSet(set, []netip.Prefix{netip.MustParsePrefix("192.168.1.0/24")})
 	require.NoError(t, err)
 
 	// Now the packet should be allowed
-	_, isAllowed = manager.routeACLsPass(srcIP, dstIP, fw.ProtocolTCP, 12345, 80)
+	_, isAllowed = manager.routeACLsPass(srcIP, dstIP, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
 	require.True(t, isAllowed, "After set update, traffic to the added network should be allowed")
 }
+
+// TestRouteACLFilteringIPv6 tests IPv6 route ACL matching directly via routeACLsPass.
+// Note: full FilterInbound for routed IPv6 traffic drops at the forwarder stage (IPv4-only)
+// but the ACL decision itself is correct.
+func TestRouteACLFilteringIPv6(t *testing.T) {
+	manager := setupRoutedManager(t, "10.10.0.100/16")
+
+	v6Dst := netip.MustParsePrefix("fd00:dead:beef::/48")
+	_, err := manager.AddRouteFiltering(
+		nil,
+		[]netip.Prefix{netip.MustParsePrefix("fd00::/16")},
+		fw.Network{Prefix: v6Dst},
+		fw.ProtocolTCP,
+		nil,
+		&fw.Port{Values: []uint16{80}},
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+
+	_, err = manager.AddRouteFiltering(
+		nil,
+		[]netip.Prefix{netip.MustParsePrefix("fd00::/16")},
+		fw.Network{Prefix: netip.MustParsePrefix("fd00:dead:beef:1::/64")},
+		fw.ProtocolALL,
+		nil,
+		nil,
+		fw.ActionDrop,
+	)
+	require.NoError(t, err)
+
+	tests := []struct {
+		name    string
+		srcIP   netip.Addr
+		dstIP   netip.Addr
+		proto   gopacket.LayerType
+		srcPort uint16
+		dstPort uint16
+		allowed bool
+	}{
+		{
+			name:    "IPv6 TCP to allowed dest",
+			srcIP:   netip.MustParseAddr("fd00::1"),
+			dstIP:   netip.MustParseAddr("fd00:dead:beef::80"),
+			proto:   layers.LayerTypeTCP,
+			srcPort: 12345,
+			dstPort: 80,
+			allowed: true,
+		},
+		{
+			name:    "IPv6 TCP wrong port",
+			srcIP:   netip.MustParseAddr("fd00::1"),
+			dstIP:   netip.MustParseAddr("fd00:dead:beef::80"),
+			proto:   layers.LayerTypeTCP,
+			srcPort: 12345,
+			dstPort: 443,
+			allowed: false,
+		},
+		{
+			name:    "IPv6 UDP not matched by TCP rule",
+			srcIP:   netip.MustParseAddr("fd00::1"),
+			dstIP:   netip.MustParseAddr("fd00:dead:beef::80"),
+			proto:   layers.LayerTypeUDP,
+			srcPort: 12345,
+			dstPort: 80,
+			allowed: false,
+		},
+		{
+			name:    "IPv6 ICMPv6 matches ICMP rule via protoLayerMatches",
+			srcIP:   netip.MustParseAddr("fd00::1"),
+			dstIP:   netip.MustParseAddr("fd00:dead:beef::80"),
+			proto:   layers.LayerTypeICMPv6,
+			allowed: false,
+		},
+		{
+			name:    "IPv6 to denied subnet",
+			srcIP:   netip.MustParseAddr("fd00::1"),
+			dstIP:   netip.MustParseAddr("fd00:dead:beef:1::1"),
+			proto:   layers.LayerTypeTCP,
+			srcPort: 12345,
+			dstPort: 80,
+			allowed: false,
+		},
+		{
+			name:    "IPv6 source outside allowed range",
+			srcIP:   netip.MustParseAddr("fe80::1"),
+			dstIP:   netip.MustParseAddr("fd00:dead:beef::80"),
+			proto:   layers.LayerTypeTCP,
+			srcPort: 12345,
+			dstPort: 80,
+			allowed: false,
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			_, pass := manager.routeACLsPass(tc.srcIP, tc.dstIP, tc.proto, tc.srcPort, tc.dstPort)
+			require.Equal(t, tc.allowed, pass, "route ACL result mismatch")
+		})
+	}
+}

client/firewall/uspfilter/filter_routeacl_test.go

@@ -0,0 +1,376 @@
+package uspfilter
+
+import (
+	"net/netip"
+	"testing"
+
+	"github.com/golang/mock/gomock"
+	"github.com/google/gopacket/layers"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	wgdevice "golang.zx2c4.com/wireguard/device"
+
+	fw "github.com/netbirdio/netbird/client/firewall/manager"
+	"github.com/netbirdio/netbird/client/iface"
+	"github.com/netbirdio/netbird/client/iface/device"
+	"github.com/netbirdio/netbird/client/iface/mocks"
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+)
+
+// TestAddRouteFilteringReturnsExistingRule verifies that adding the same route
+// filtering rule twice returns the same rule ID (idempotent behavior).
+func TestAddRouteFilteringReturnsExistingRule(t *testing.T) {
+	manager := setupTestManager(t)
+
+	sources := []netip.Prefix{
+		netip.MustParsePrefix("100.64.1.0/24"),
+		netip.MustParsePrefix("100.64.2.0/24"),
+	}
+	destination := fw.Network{Prefix: netip.MustParsePrefix("192.168.1.0/24")}
+
+	// Add rule first time
+	rule1, err := manager.AddRouteFiltering(
+		[]byte("policy-1"),
+		sources,
+		destination,
+		fw.ProtocolTCP,
+		nil,
+		&fw.Port{Values: []uint16{443}},
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+	require.NotNil(t, rule1)
+
+	// Add the same rule again
+	rule2, err := manager.AddRouteFiltering(
+		[]byte("policy-1"),
+		sources,
+		destination,
+		fw.ProtocolTCP,
+		nil,
+		&fw.Port{Values: []uint16{443}},
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+	require.NotNil(t, rule2)
+
+	// These should be the same (idempotent) like nftables/iptables implementations
+	assert.Equal(t, rule1.ID(), rule2.ID(),
+		"Adding the same rule twice should return the same rule ID (idempotent)")
+
+	manager.mutex.RLock()
+	ruleCount := len(manager.routeRules)
+	manager.mutex.RUnlock()
+
+	assert.Equal(t, 2, ruleCount,
+		"Should have exactly 2 rules (1 user rule + 1 block rule)")
+}
+
+// TestAddRouteFilteringDifferentRulesGetDifferentIDs verifies that rules with
+// different parameters get distinct IDs.
+func TestAddRouteFilteringDifferentRulesGetDifferentIDs(t *testing.T) {
+	manager := setupTestManager(t)
+
+	sources := []netip.Prefix{netip.MustParsePrefix("100.64.1.0/24")}
+
+	// Add first rule
+	rule1, err := manager.AddRouteFiltering(
+		[]byte("policy-1"),
+		sources,
+		fw.Network{Prefix: netip.MustParsePrefix("192.168.1.0/24")},
+		fw.ProtocolTCP,
+		nil,
+		&fw.Port{Values: []uint16{443}},
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+
+	// Add different rule (different destination)
+	rule2, err := manager.AddRouteFiltering(
+		[]byte("policy-2"),
+		sources,
+		fw.Network{Prefix: netip.MustParsePrefix("192.168.2.0/24")}, // Different!
+		fw.ProtocolTCP,
+		nil,
+		&fw.Port{Values: []uint16{443}},
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+
+	assert.NotEqual(t, rule1.ID(), rule2.ID(),
+		"Different rules should have different IDs")
+
+	manager.mutex.RLock()
+	ruleCount := len(manager.routeRules)
+	manager.mutex.RUnlock()
+
+	assert.Equal(t, 3, ruleCount, "Should have 3 rules (2 user rules + 1 block rule)")
+}
+
+// TestRouteRuleUpdateDoesNotCauseGap verifies that re-adding the same route
+// rule during a network map update does not disrupt existing traffic.
+func TestRouteRuleUpdateDoesNotCauseGap(t *testing.T) {
+	manager := setupTestManager(t)
+
+	sources := []netip.Prefix{netip.MustParsePrefix("100.64.1.0/24")}
+	destination := fw.Network{Prefix: netip.MustParsePrefix("192.168.1.0/24")}
+
+	rule1, err := manager.AddRouteFiltering(
+		[]byte("policy-1"),
+		sources,
+		destination,
+		fw.ProtocolTCP,
+		nil,
+		nil,
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+
+	srcIP := netip.MustParseAddr("100.64.1.5")
+	dstIP := netip.MustParseAddr("192.168.1.10")
+	_, pass := manager.routeACLsPass(srcIP, dstIP, layers.LayerTypeTCP, 12345, 443)
+	require.True(t, pass, "Traffic should pass with rule in place")
+
+	// Re-add same rule (simulates network map update)
+	rule2, err := manager.AddRouteFiltering(
+		[]byte("policy-1"),
+		sources,
+		destination,
+		fw.ProtocolTCP,
+		nil,
+		nil,
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+
+	// Idempotent IDs mean rule1.ID() == rule2.ID(), so the ACL manager
+	// won't delete rule1 during cleanup. If IDs differed, deleting rule1
+	// would remove the only matching rule and cause a traffic gap.
+	if rule1.ID() != rule2.ID() {
+		err = manager.DeleteRouteRule(rule1)
+		require.NoError(t, err)
+	}
+
+	_, passAfter := manager.routeACLsPass(srcIP, dstIP, layers.LayerTypeTCP, 12345, 443)
+	assert.True(t, passAfter,
+		"Traffic should still pass after rule update - no gap should occur")
+}
+
+// TestBlockInvalidRoutedIdempotent verifies that blockInvalidRouted creates
+// exactly one drop rule for the WireGuard network prefix, and calling it again
+// returns the same rule without duplicating.
+func TestBlockInvalidRoutedIdempotent(t *testing.T) {
+	ctrl := gomock.NewController(t)
+	dev := mocks.NewMockDevice(ctrl)
+	dev.EXPECT().MTU().Return(1500, nil).AnyTimes()
+
+	wgNet := netip.MustParsePrefix("100.64.0.1/16")
+
+	ifaceMock := &IFaceMock{
+		SetFilterFunc: func(device.PacketFilter) error { return nil },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      wgNet.Addr(),
+				Network: wgNet,
+			}
+		},
+		GetDeviceFunc: func() *device.FilteredDevice {
+			return &device.FilteredDevice{Device: dev}
+		},
+		GetWGDeviceFunc: func() *wgdevice.Device {
+			return &wgdevice.Device{}
+		},
+	}
+
+	manager, err := Create(ifaceMock, false, flowLogger, iface.DefaultMTU)
+	require.NoError(t, err)
+	t.Cleanup(func() {
+		require.NoError(t, manager.Close(nil))
+	})
+
+	// Call blockInvalidRouted directly multiple times
+	rule1, err := manager.blockInvalidRouted(ifaceMock)
+	require.NoError(t, err)
+	require.NotNil(t, rule1)
+
+	rule2, err := manager.blockInvalidRouted(ifaceMock)
+	require.NoError(t, err)
+	require.NotNil(t, rule2)
+
+	rule3, err := manager.blockInvalidRouted(ifaceMock)
+	require.NoError(t, err)
+	require.NotNil(t, rule3)
+
+	// All should return the same rule
+	assert.Equal(t, rule1.ID(), rule2.ID(), "Second call should return same rule")
+	assert.Equal(t, rule2.ID(), rule3.ID(), "Third call should return same rule")
+
+	// Should have exactly 1 route rule
+	manager.mutex.RLock()
+	ruleCount := len(manager.routeRules)
+	manager.mutex.RUnlock()
+
+	assert.Equal(t, 1, ruleCount, "Should have exactly 1 block rule after 3 calls")
+
+	// Verify the rule blocks traffic to the WG network
+	srcIP := netip.MustParseAddr("10.0.0.1")
+	dstIP := netip.MustParseAddr("100.64.0.50")
+	_, pass := manager.routeACLsPass(srcIP, dstIP, layers.LayerTypeTCP, 12345, 80)
+	assert.False(t, pass, "Block rule should deny traffic to WG prefix")
+}
+
+// TestBlockRuleNotAccumulatedOnRepeatedEnableRouting verifies that calling
+// EnableRouting multiple times (as happens on each route update) does not
+// accumulate duplicate block rules in the routeRules slice.
+func TestBlockRuleNotAccumulatedOnRepeatedEnableRouting(t *testing.T) {
+	ctrl := gomock.NewController(t)
+	dev := mocks.NewMockDevice(ctrl)
+	dev.EXPECT().MTU().Return(1500, nil).AnyTimes()
+
+	wgNet := netip.MustParsePrefix("100.64.0.1/16")
+
+	ifaceMock := &IFaceMock{
+		SetFilterFunc: func(device.PacketFilter) error { return nil },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      wgNet.Addr(),
+				Network: wgNet,
+			}
+		},
+		GetDeviceFunc: func() *device.FilteredDevice {
+			return &device.FilteredDevice{Device: dev}
+		},
+		GetWGDeviceFunc: func() *wgdevice.Device {
+			return &wgdevice.Device{}
+		},
+	}
+
+	manager, err := Create(ifaceMock, false, flowLogger, iface.DefaultMTU)
+	require.NoError(t, err)
+	t.Cleanup(func() {
+		require.NoError(t, manager.Close(nil))
+	})
+
+	// Call EnableRouting multiple times (simulating repeated route updates)
+	for i := 0; i < 5; i++ {
+		require.NoError(t, manager.EnableRouting())
+	}
+
+	manager.mutex.RLock()
+	ruleCount := len(manager.routeRules)
+	manager.mutex.RUnlock()
+
+	assert.Equal(t, 1, ruleCount,
+		"Repeated EnableRouting should not accumulate block rules")
+}
+
+// TestRouteRuleCountStableAcrossUpdates verifies that adding the same route
+// rule multiple times does not create duplicate entries.
+func TestRouteRuleCountStableAcrossUpdates(t *testing.T) {
+	manager := setupTestManager(t)
+
+	sources := []netip.Prefix{netip.MustParsePrefix("100.64.1.0/24")}
+	destination := fw.Network{Prefix: netip.MustParsePrefix("192.168.1.0/24")}
+
+	// Simulate 5 network map updates with the same route rule
+	for i := 0; i < 5; i++ {
+		rule, err := manager.AddRouteFiltering(
+			[]byte("policy-1"),
+			sources,
+			destination,
+			fw.ProtocolTCP,
+			nil,
+			&fw.Port{Values: []uint16{443}},
+			fw.ActionAccept,
+		)
+		require.NoError(t, err)
+		require.NotNil(t, rule)
+	}
+
+	manager.mutex.RLock()
+	ruleCount := len(manager.routeRules)
+	manager.mutex.RUnlock()
+
+	assert.Equal(t, 2, ruleCount,
+		"Should have exactly 2 rules (1 user rule + 1 block rule) after 5 updates")
+}
+
+// TestDeleteRouteRuleAfterIdempotentAdd verifies that deleting a route rule
+// after adding it multiple times works correctly.
+func TestDeleteRouteRuleAfterIdempotentAdd(t *testing.T) {
+	manager := setupTestManager(t)
+
+	sources := []netip.Prefix{netip.MustParsePrefix("100.64.1.0/24")}
+	destination := fw.Network{Prefix: netip.MustParsePrefix("192.168.1.0/24")}
+
+	// Add same rule twice
+	rule1, err := manager.AddRouteFiltering(
+		[]byte("policy-1"),
+		sources,
+		destination,
+		fw.ProtocolTCP,
+		nil,
+		nil,
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+
+	rule2, err := manager.AddRouteFiltering(
+		[]byte("policy-1"),
+		sources,
+		destination,
+		fw.ProtocolTCP,
+		nil,
+		nil,
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+
+	require.Equal(t, rule1.ID(), rule2.ID(), "Should return same rule ID")
+
+	// Delete using first reference
+	err = manager.DeleteRouteRule(rule1)
+	require.NoError(t, err)
+
+	// Verify traffic no longer passes
+	srcIP := netip.MustParseAddr("100.64.1.5")
+	dstIP := netip.MustParseAddr("192.168.1.10")
+	_, pass := manager.routeACLsPass(srcIP, dstIP, layers.LayerTypeTCP, 12345, 443)
+	assert.False(t, pass, "Traffic should not pass after rule deletion")
+}
+
+func setupTestManager(t *testing.T) *Manager {
+	t.Helper()
+
+	ctrl := gomock.NewController(t)
+	dev := mocks.NewMockDevice(ctrl)
+	dev.EXPECT().MTU().Return(1500, nil).AnyTimes()
+
+	wgNet := netip.MustParsePrefix("100.64.0.1/16")
+
+	ifaceMock := &IFaceMock{
+		SetFilterFunc: func(device.PacketFilter) error { return nil },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      wgNet.Addr(),
+				Network: wgNet,
+			}
+		},
+		GetDeviceFunc: func() *device.FilteredDevice {
+			return &device.FilteredDevice{Device: dev}
+		},
+		GetWGDeviceFunc: func() *wgdevice.Device {
+			return &wgdevice.Device{}
+		},
+	}
+
+	manager, err := Create(ifaceMock, false, flowLogger, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, manager.EnableRouting())
+
+	t.Cleanup(func() {
+		require.NoError(t, manager.Close(nil))
+	})
+
+	return manager
+}

client/firewall/uspfilter/filter_test.go

@@ -263,6 +263,158 @@ func TestAddUDPPacketHook(t *testing.T) {
 	}
 }
 
+// TestPeerRuleLifecycleDenyRules verifies that deny rules are correctly added
+// to the deny map and can be cleanly deleted without leaving orphans.
+func TestPeerRuleLifecycleDenyRules(t *testing.T) {
+	ifaceMock := &IFaceMock{
+		SetFilterFunc: func(device.PacketFilter) error { return nil },
+	}
+
+	m, err := Create(ifaceMock, false, flowLogger, nbiface.DefaultMTU)
+	require.NoError(t, err)
+	defer func() {
+		require.NoError(t, m.Close(nil))
+	}()
+
+	ip := net.ParseIP("192.168.1.1")
+	addr := netip.MustParseAddr("192.168.1.1")
+
+	// Add multiple deny rules for different ports
+	rule1, err := m.AddPeerFiltering(nil, ip, fw.ProtocolTCP, nil,
+		&fw.Port{Values: []uint16{22}}, fw.ActionDrop, "")
+	require.NoError(t, err)
+
+	rule2, err := m.AddPeerFiltering(nil, ip, fw.ProtocolTCP, nil,
+		&fw.Port{Values: []uint16{80}}, fw.ActionDrop, "")
+	require.NoError(t, err)
+
+	m.mutex.RLock()
+	denyCount := len(m.incomingDenyRules[addr])
+	m.mutex.RUnlock()
+	require.Equal(t, 2, denyCount, "Should have exactly 2 deny rules")
+
+	// Delete the first deny rule
+	err = m.DeletePeerRule(rule1[0])
+	require.NoError(t, err)
+
+	m.mutex.RLock()
+	denyCount = len(m.incomingDenyRules[addr])
+	m.mutex.RUnlock()
+	require.Equal(t, 1, denyCount, "Should have 1 deny rule after deleting first")
+
+	// Delete the second deny rule
+	err = m.DeletePeerRule(rule2[0])
+	require.NoError(t, err)
+
+	m.mutex.RLock()
+	_, exists := m.incomingDenyRules[addr]
+	m.mutex.RUnlock()
+	require.False(t, exists, "Deny rules IP entry should be cleaned up when empty")
+}
+
+// TestPeerRuleAddAndDeleteDontLeak verifies that repeatedly adding and deleting
+// peer rules (simulating network map updates) does not leak rules in the maps.
+func TestPeerRuleAddAndDeleteDontLeak(t *testing.T) {
+	ifaceMock := &IFaceMock{
+		SetFilterFunc: func(device.PacketFilter) error { return nil },
+	}
+
+	m, err := Create(ifaceMock, false, flowLogger, nbiface.DefaultMTU)
+	require.NoError(t, err)
+	defer func() {
+		require.NoError(t, m.Close(nil))
+	}()
+
+	ip := net.ParseIP("192.168.1.1")
+	addr := netip.MustParseAddr("192.168.1.1")
+
+	// Simulate 10 network map updates: add rule, delete old, add new
+	for i := 0; i < 10; i++ {
+		// Add a deny rule
+		rules, err := m.AddPeerFiltering(nil, ip, fw.ProtocolTCP, nil,
+			&fw.Port{Values: []uint16{22}}, fw.ActionDrop, "")
+		require.NoError(t, err)
+
+		// Add an allow rule
+		allowRules, err := m.AddPeerFiltering(nil, ip, fw.ProtocolTCP, nil,
+			&fw.Port{Values: []uint16{80}}, fw.ActionAccept, "")
+		require.NoError(t, err)
+
+		// Delete them (simulating ACL manager cleanup)
+		for _, r := range rules {
+			require.NoError(t, m.DeletePeerRule(r))
+		}
+		for _, r := range allowRules {
+			require.NoError(t, m.DeletePeerRule(r))
+		}
+	}
+
+	m.mutex.RLock()
+	denyCount := len(m.incomingDenyRules[addr])
+	allowCount := len(m.incomingRules[addr])
+	m.mutex.RUnlock()
+
+	require.Equal(t, 0, denyCount, "No deny rules should remain after cleanup")
+	require.Equal(t, 0, allowCount, "No allow rules should remain after cleanup")
+}
+
+// TestMixedAllowDenyRulesSameIP verifies that allow and deny rules for the same
+// IP are stored in separate maps and don't interfere with each other.
+func TestMixedAllowDenyRulesSameIP(t *testing.T) {
+	ifaceMock := &IFaceMock{
+		SetFilterFunc: func(device.PacketFilter) error { return nil },
+	}
+
+	m, err := Create(ifaceMock, false, flowLogger, nbiface.DefaultMTU)
+	require.NoError(t, err)
+	defer func() {
+		require.NoError(t, m.Close(nil))
+	}()
+
+	ip := net.ParseIP("192.168.1.1")
+
+	// Add allow rule for port 80
+	allowRule, err := m.AddPeerFiltering(nil, ip, fw.ProtocolTCP, nil,
+		&fw.Port{Values: []uint16{80}}, fw.ActionAccept, "")
+	require.NoError(t, err)
+
+	// Add deny rule for port 22
+	denyRule, err := m.AddPeerFiltering(nil, ip, fw.ProtocolTCP, nil,
+		&fw.Port{Values: []uint16{22}}, fw.ActionDrop, "")
+	require.NoError(t, err)
+
+	addr := netip.MustParseAddr("192.168.1.1")
+	m.mutex.RLock()
+	allowCount := len(m.incomingRules[addr])
+	denyCount := len(m.incomingDenyRules[addr])
+	m.mutex.RUnlock()
+
+	require.Equal(t, 1, allowCount, "Should have 1 allow rule")
+	require.Equal(t, 1, denyCount, "Should have 1 deny rule")
+
+	// Delete allow rule should not affect deny rule
+	err = m.DeletePeerRule(allowRule[0])
+	require.NoError(t, err)
+
+	m.mutex.RLock()
+	denyCountAfter := len(m.incomingDenyRules[addr])
+	m.mutex.RUnlock()
+
+	require.Equal(t, 1, denyCountAfter, "Deny rule should still exist after deleting allow rule")
+
+	// Delete deny rule
+	err = m.DeletePeerRule(denyRule[0])
+	require.NoError(t, err)
+
+	m.mutex.RLock()
+	_, denyExists := m.incomingDenyRules[addr]
+	_, allowExists := m.incomingRules[addr]
+	m.mutex.RUnlock()
+
+	require.False(t, denyExists, "Deny rules should be empty")
+	require.False(t, allowExists, "Allow rules should be empty")
+}
+
 func TestManagerReset(t *testing.T) {
 	ifaceMock := &IFaceMock{
 		SetFilterFunc: func(device.PacketFilter) error { return nil },
@@ -430,11 +582,16 @@ func TestProcessOutgoingHooks(t *testing.T) {
 			d := &decoder{
 				decoded: []gopacket.LayerType{},
 			}
-			d.parser = gopacket.NewDecodingLayerParser(
+			d.parser4 = gopacket.NewDecodingLayerParser(
 				layers.LayerTypeIPv4,
 				&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
 			)
-			d.parser.IgnoreUnsupported = true
+			d.parser4.IgnoreUnsupported = true
+			d.parser6 = gopacket.NewDecodingLayerParser(
+				layers.LayerTypeIPv6,
+				&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
+			)
+			d.parser6.IgnoreUnsupported = true
 			return d
 		},
 	}
@@ -535,11 +692,16 @@ func TestStatefulFirewall_UDPTracking(t *testing.T) {
 			d := &decoder{
 				decoded: []gopacket.LayerType{},
 			}
-			d.parser = gopacket.NewDecodingLayerParser(
+			d.parser4 = gopacket.NewDecodingLayerParser(
 				layers.LayerTypeIPv4,
 				&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
 			)
-			d.parser.IgnoreUnsupported = true
+			d.parser4.IgnoreUnsupported = true
+			d.parser6 = gopacket.NewDecodingLayerParser(
+				layers.LayerTypeIPv6,
+				&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
+			)
+			d.parser6.IgnoreUnsupported = true
 			return d
 		},
 	}
@@ -767,9 +929,9 @@ func TestUpdateSetMerge(t *testing.T) {
 	dstIP2 := netip.MustParseAddr("192.168.1.100")
 	dstIP3 := netip.MustParseAddr("172.16.0.100")
 
-	_, isAllowed1 := manager.routeACLsPass(srcIP, dstIP1, fw.ProtocolTCP, 12345, 80)
-	_, isAllowed2 := manager.routeACLsPass(srcIP, dstIP2, fw.ProtocolTCP, 12345, 80)
-	_, isAllowed3 := manager.routeACLsPass(srcIP, dstIP3, fw.ProtocolTCP, 12345, 80)
+	_, isAllowed1 := manager.routeACLsPass(srcIP, dstIP1, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
+	_, isAllowed2 := manager.routeACLsPass(srcIP, dstIP2, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
+	_, isAllowed3 := manager.routeACLsPass(srcIP, dstIP3, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
 
 	require.True(t, isAllowed1, "Traffic to 10.0.0.100 should be allowed")
 	require.True(t, isAllowed2, "Traffic to 192.168.1.100 should be allowed")
@@ -784,8 +946,8 @@ func TestUpdateSetMerge(t *testing.T) {
 	require.NoError(t, err)
 
 	// Check that all original prefixes are still included
-	_, isAllowed1 = manager.routeACLsPass(srcIP, dstIP1, fw.ProtocolTCP, 12345, 80)
-	_, isAllowed2 = manager.routeACLsPass(srcIP, dstIP2, fw.ProtocolTCP, 12345, 80)
+	_, isAllowed1 = manager.routeACLsPass(srcIP, dstIP1, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
+	_, isAllowed2 = manager.routeACLsPass(srcIP, dstIP2, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
 	require.True(t, isAllowed1, "Traffic to 10.0.0.100 should still be allowed after update")
 	require.True(t, isAllowed2, "Traffic to 192.168.1.100 should still be allowed after update")
 
@@ -793,8 +955,8 @@ func TestUpdateSetMerge(t *testing.T) {
 	dstIP4 := netip.MustParseAddr("172.16.1.100")
 	dstIP5 := netip.MustParseAddr("10.1.0.50")
 
-	_, isAllowed4 := manager.routeACLsPass(srcIP, dstIP4, fw.ProtocolTCP, 12345, 80)
-	_, isAllowed5 := manager.routeACLsPass(srcIP, dstIP5, fw.ProtocolTCP, 12345, 80)
+	_, isAllowed4 := manager.routeACLsPass(srcIP, dstIP4, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
+	_, isAllowed5 := manager.routeACLsPass(srcIP, dstIP5, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
 
 	require.True(t, isAllowed4, "Traffic to new prefix 172.16.0.0/16 should be allowed")
 	require.True(t, isAllowed5, "Traffic to new prefix 10.1.0.0/24 should be allowed")
@@ -922,7 +1084,7 @@ func TestUpdateSetDeduplication(t *testing.T) {
 
 	srcIP := netip.MustParseAddr("100.10.0.1")
 	for _, tc := range testCases {
-		_, isAllowed := manager.routeACLsPass(srcIP, tc.dstIP, fw.ProtocolTCP, 12345, 80)
+		_, isAllowed := manager.routeACLsPass(srcIP, tc.dstIP, protoToLayer(fw.ProtocolTCP, layers.LayerTypeIPv4), 12345, 80)
 		require.Equal(t, tc.expected, isAllowed, tc.desc)
 	}
 }
@@ -945,8 +1107,8 @@ func TestMSSClamping(t *testing.T) {
 	}()
 
 	require.True(t, manager.mssClampEnabled, "MSS clamping should be enabled by default")
-	expectedMSSValue := uint16(1280 - ipTCPHeaderMinSize)
-	require.Equal(t, expectedMSSValue, manager.mssClampValue, "MSS clamp value should be MTU - 40")
+	require.Equal(t, uint16(1280-ipv4TCPHeaderMinSize), manager.mssClampValueIPv4, "IPv4 MSS clamp value should be MTU - 40")
+	require.Equal(t, uint16(1280-ipv6TCPHeaderMinSize), manager.mssClampValueIPv6, "IPv6 MSS clamp value should be MTU - 60")
 
 	err = manager.UpdateLocalIPs()
 	require.NoError(t, err)
@@ -964,7 +1126,7 @@ func TestMSSClamping(t *testing.T) {
 		require.Len(t, d.tcp.Options, 1, "Should have MSS option")
 		require.Equal(t, uint8(layers.TCPOptionKindMSS), uint8(d.tcp.Options[0].OptionType))
 		actualMSS := binary.BigEndian.Uint16(d.tcp.Options[0].OptionData)
-		require.Equal(t, expectedMSSValue, actualMSS, "MSS should be clamped to MTU - 40")
+		require.Equal(t, manager.mssClampValueIPv4, actualMSS, "MSS should be clamped to MTU - 40")
 	})
 
 	t.Run("SYN packet with low MSS unchanged", func(t *testing.T) {
@@ -988,7 +1150,7 @@ func TestMSSClamping(t *testing.T) {
 		d := parsePacket(t, packet)
 		require.Len(t, d.tcp.Options, 1, "Should have MSS option")
 		actualMSS := binary.BigEndian.Uint16(d.tcp.Options[0].OptionData)
-		require.Equal(t, expectedMSSValue, actualMSS, "MSS in SYN-ACK should be clamped")
+		require.Equal(t, manager.mssClampValueIPv4, actualMSS, "MSS in SYN-ACK should be clamped")
 	})
 
 	t.Run("Non-SYN packet unchanged", func(t *testing.T) {
@@ -1160,13 +1322,18 @@ func TestShouldForward(t *testing.T) {
 		d := &decoder{
 			decoded: []gopacket.LayerType{},
 		}
-		d.parser = gopacket.NewDecodingLayerParser(
+		d.parser4 = gopacket.NewDecodingLayerParser(
 			layers.LayerTypeIPv4,
 			&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
 		)
-		d.parser.IgnoreUnsupported = true
+		d.parser4.IgnoreUnsupported = true
+		d.parser6 = gopacket.NewDecodingLayerParser(
+			layers.LayerTypeIPv6,
+			&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
+		)
+		d.parser6.IgnoreUnsupported = true
 
-		err = d.parser.DecodeLayers(buf.Bytes(), &d.decoded)
+		err = d.decodePacket(buf.Bytes())
 		require.NoError(t, err)
 
 		return d
@@ -1226,6 +1393,44 @@ func TestShouldForward(t *testing.T) {
 		},
 	}
 
+	// Add IPv6 to the interface and test dual-stack cases
+	wgIPv6 := netip.MustParseAddr("fd00::1")
+	otherIPv6 := netip.MustParseAddr("fd00::2")
+	ifaceMock.AddressFunc = func() wgaddr.Address {
+		return wgaddr.Address{
+			IP:      wgIP,
+			Network: netip.PrefixFrom(wgIP, 24),
+			IPv6:    wgIPv6,
+			IPv6Net: netip.PrefixFrom(wgIPv6, 64),
+		}
+	}
+
+	// Re-create manager to pick up the new address with IPv6
+	require.NoError(t, manager.Close(nil))
+	manager, err = Create(ifaceMock, false, flowLogger, nbiface.DefaultMTU)
+	require.NoError(t, err)
+
+	v6Cases := []struct {
+		name        string
+		dstIP       netip.Addr
+		expected    bool
+		description string
+	}{
+		{"v6 traffic to other address", otherIPv6, true, "should forward v6 traffic not destined to our v6 address"},
+		{"v6 traffic to our v6 IP", wgIPv6, false, "should not forward traffic destined to our v6 address"},
+		{"v4 traffic to other with v6 configured", otherIP, true, "should forward v4 traffic when v6 configured"},
+		{"v4 traffic to our v4 IP with v6 configured", wgIP, false, "should not forward traffic to our v4 address"},
+	}
+	for _, tt := range v6Cases {
+		t.Run(tt.name, func(t *testing.T) {
+			manager.localForwarding = true
+			manager.netstack = false
+			decoder := createTCPDecoder(8080)
+			result := manager.shouldForward(decoder, tt.dstIP)
+			require.Equal(t, tt.expected, result, tt.description)
+		})
+	}
+
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			// Configure manager

client/firewall/uspfilter/forwarder/endpoint.go

@@ -1,7 +1,9 @@
 package forwarder
 
 import (
-	"fmt"
+	"net"
+	"strconv"
+	"sync/atomic"
 
 	wgdevice "golang.zx2c4.com/wireguard/device"
 	"gvisor.dev/gvisor/pkg/tcpip"
@@ -16,7 +18,7 @@ type endpoint struct {
 	logger     *nblog.Logger
 	dispatcher stack.NetworkDispatcher
 	device     *wgdevice.Device
-	mtu        uint32
+	mtu        atomic.Uint32
 }
 
 func (e *endpoint) Attach(dispatcher stack.NetworkDispatcher) {
@@ -28,7 +30,7 @@ func (e *endpoint) IsAttached() bool {
 }
 
 func (e *endpoint) MTU() uint32 {
-	return e.mtu
+	return e.mtu.Load()
 }
 
 func (e *endpoint) Capabilities() stack.LinkEndpointCapabilities {
@@ -46,17 +48,23 @@ func (e *endpoint) LinkAddress() tcpip.LinkAddress {
 func (e *endpoint) WritePackets(pkts stack.PacketBufferList) (int, tcpip.Error) {
 	var written int
 	for _, pkt := range pkts.AsSlice() {
-		netHeader := header.IPv4(pkt.NetworkHeader().View().AsSlice())
-
 		data := stack.PayloadSince(pkt.NetworkHeader())
 		if data == nil {
 			continue
 		}
 
-		// Send the packet through WireGuard
-		address := netHeader.DestinationAddress()
-		err := e.device.CreateOutboundPacket(data.AsSlice(), address.AsSlice())
-		if err != nil {
+		raw := pkt.NetworkHeader().View().AsSlice()
+		if len(raw) == 0 {
+			continue
+		}
+		var address tcpip.Address
+		if raw[0]>>4 == 6 {
+			address = header.IPv6(raw).DestinationAddress()
+		} else {
+			address = header.IPv4(raw).DestinationAddress()
+		}
+
+		if err := e.device.CreateOutboundPacket(data.AsSlice(), address.AsSlice()); err != nil {
 			e.logger.Error1("CreateOutboundPacket: %v", err)
 			continue
 		}
@@ -82,9 +90,27 @@ func (e *endpoint) ParseHeader(*stack.PacketBuffer) bool {
 	return true
 }
 
+func (e *endpoint) Close() {
+	// Endpoint cleanup - nothing to do as device is managed externally
+}
+
+func (e *endpoint) SetLinkAddress(tcpip.LinkAddress) {
+	// Link address is not used for this endpoint type
+}
+
+func (e *endpoint) SetMTU(mtu uint32) {
+	e.mtu.Store(mtu)
+}
+
+func (e *endpoint) SetOnCloseAction(func()) {
+	// No action needed on close
+}
+
 type epID stack.TransportEndpointID
 
 func (i epID) String() string {
 	// src and remote is swapped
-	return fmt.Sprintf("%s:%d → %s:%d", i.RemoteAddress, i.RemotePort, i.LocalAddress, i.LocalPort)
+	return net.JoinHostPort(i.RemoteAddress.String(), strconv.Itoa(int(i.RemotePort))) +
+		" → " +
+		net.JoinHostPort(i.LocalAddress.String(), strconv.Itoa(int(i.LocalPort)))
 }

client/firewall/uspfilter/forwarder/forwarder.go

@@ -7,12 +7,14 @@ import (
 	"net/netip"
 	"runtime"
 	"sync"
+	"time"
 
 	log "github.com/sirupsen/logrus"
 	"gvisor.dev/gvisor/pkg/buffer"
 	"gvisor.dev/gvisor/pkg/tcpip"
 	"gvisor.dev/gvisor/pkg/tcpip/header"
 	"gvisor.dev/gvisor/pkg/tcpip/network/ipv4"
+	"gvisor.dev/gvisor/pkg/tcpip/network/ipv6"
 	"gvisor.dev/gvisor/pkg/tcpip/stack"
 	"gvisor.dev/gvisor/pkg/tcpip/transport/icmp"
 	"gvisor.dev/gvisor/pkg/tcpip/transport/tcp"
@@ -35,23 +37,31 @@ type Forwarder struct {
 	logger     *nblog.Logger
 	flowLogger nftypes.FlowLogger
 	// ruleIdMap is used to store the rule ID for a given connection
-	ruleIdMap    sync.Map
-	stack        *stack.Stack
-	endpoint     *endpoint
-	udpForwarder *udpForwarder
-	ctx          context.Context
-	cancel       context.CancelFunc
-	ip           tcpip.Address
-	netstack     bool
+	ruleIdMap          sync.Map
+	stack              *stack.Stack
+	endpoint           *endpoint
+	udpForwarder       *udpForwarder
+	ctx                context.Context
+	cancel             context.CancelFunc
+	ip                 tcpip.Address
+	ipv6               tcpip.Address
+	netstack           bool
+	hasRawICMPAccess   bool
+	hasRawICMPv6Access bool
+	pingSemaphore      chan struct{}
 }
 
 func New(iface common.IFaceMapper, logger *nblog.Logger, flowLogger nftypes.FlowLogger, netstack bool, mtu uint16) (*Forwarder, error) {
 	s := stack.New(stack.Options{
-		NetworkProtocols: []stack.NetworkProtocolFactory{ipv4.NewProtocol},
+		NetworkProtocols: []stack.NetworkProtocolFactory{
+			ipv4.NewProtocol,
+			ipv6.NewProtocol,
+		},
 		TransportProtocols: []stack.TransportProtocolFactory{
 			tcp.NewProtocol,
 			udp.NewProtocol,
 			icmp.NewProtocol4,
+			icmp.NewProtocol6,
 		},
 		HandleLocal: false,
 	})
@@ -60,8 +70,8 @@ func New(iface common.IFaceMapper, logger *nblog.Logger, flowLogger nftypes.Flow
 	endpoint := &endpoint{
 		logger: logger,
 		device: iface.GetWGDevice(),
-		mtu:    uint32(mtu),
 	}
+	endpoint.mtu.Store(uint32(mtu))
 
 	if err := s.CreateNIC(nicID, endpoint); err != nil {
 		return nil, fmt.Errorf("create NIC: %v", err)
@@ -70,7 +80,7 @@ func New(iface common.IFaceMapper, logger *nblog.Logger, flowLogger nftypes.Flow
 	protoAddr := tcpip.ProtocolAddress{
 		Protocol: ipv4.ProtocolNumber,
 		AddressWithPrefix: tcpip.AddressWithPrefix{
-			Address:   tcpip.AddrFromSlice(iface.Address().IP.AsSlice()),
+			Address:   tcpip.AddrFrom4(iface.Address().IP.As4()),
 			PrefixLen: iface.Address().Network.Bits(),
 		},
 	}
@@ -79,6 +89,19 @@ func New(iface common.IFaceMapper, logger *nblog.Logger, flowLogger nftypes.Flow
 		return nil, fmt.Errorf("failed to add protocol address: %s", err)
 	}
 
+	if v6 := iface.Address().IPv6; v6.IsValid() {
+		v6Addr := tcpip.ProtocolAddress{
+			Protocol: ipv6.ProtocolNumber,
+			AddressWithPrefix: tcpip.AddressWithPrefix{
+				Address:   tcpip.AddrFrom16(v6.As16()),
+				PrefixLen: iface.Address().IPv6Net.Bits(),
+			},
+		}
+		if err := s.AddProtocolAddress(nicID, v6Addr, stack.AddressProperties{}); err != nil {
+			return nil, fmt.Errorf("add IPv6 protocol address: %s", err)
+		}
+	}
+
 	defaultSubnet, err := tcpip.NewSubnet(
 		tcpip.AddrFrom4([4]byte{0, 0, 0, 0}),
 		tcpip.MaskFromBytes([]byte{0, 0, 0, 0}),
@@ -87,6 +110,14 @@ func New(iface common.IFaceMapper, logger *nblog.Logger, flowLogger nftypes.Flow
 		return nil, fmt.Errorf("creating default subnet: %w", err)
 	}
 
+	defaultSubnetV6, err := tcpip.NewSubnet(
+		tcpip.AddrFrom16([16]byte{}),
+		tcpip.MaskFromBytes(make([]byte, 16)),
+	)
+	if err != nil {
+		return nil, fmt.Errorf("creating default v6 subnet: %w", err)
+	}
+
 	if err := s.SetPromiscuousMode(nicID, true); err != nil {
 		return nil, fmt.Errorf("set promiscuous mode: %s", err)
 	}
@@ -95,23 +126,23 @@ func New(iface common.IFaceMapper, logger *nblog.Logger, flowLogger nftypes.Flow
 	}
 
 	s.SetRouteTable([]tcpip.Route{
-		{
-			Destination: defaultSubnet,
-			NIC:         nicID,
-		},
+		{Destination: defaultSubnet, NIC: nicID},
+		{Destination: defaultSubnetV6, NIC: nicID},
 	})
 
 	ctx, cancel := context.WithCancel(context.Background())
 	f := &Forwarder{
-		logger:       logger,
-		flowLogger:   flowLogger,
-		stack:        s,
-		endpoint:     endpoint,
-		udpForwarder: newUDPForwarder(mtu, logger, flowLogger),
-		ctx:          ctx,
-		cancel:       cancel,
-		netstack:     netstack,
-		ip:           tcpip.AddrFromSlice(iface.Address().IP.AsSlice()),
+		logger:        logger,
+		flowLogger:    flowLogger,
+		stack:         s,
+		endpoint:      endpoint,
+		udpForwarder:  newUDPForwarder(mtu, logger, flowLogger),
+		ctx:           ctx,
+		cancel:        cancel,
+		netstack:      netstack,
+		ip:            tcpip.AddrFrom4(iface.Address().IP.As4()),
+		ipv6:          addrFromNetipAddr(iface.Address().IPv6),
+		pingSemaphore: make(chan struct{}, 3),
 	}
 
 	receiveWindow := defaultReceiveWindow
@@ -127,15 +158,42 @@ func New(iface common.IFaceMapper, logger *nblog.Logger, flowLogger nftypes.Flow
 	udpForwarder := udp.NewForwarder(s, f.handleUDP)
 	s.SetTransportProtocolHandler(udp.ProtocolNumber, udpForwarder.HandlePacket)
 
-	s.SetTransportProtocolHandler(icmp.ProtocolNumber4, f.handleICMP)
+	// ICMP is handled directly in InjectIncomingPacket, bypassing gVisor's
+	// network layer. This avoids duplicate echo replies (v4) and the v6
+	// auto-reply bug where gVisor responds at the network layer before
+	// our transport handler fires.
+
+	f.checkICMPCapability()
 
 	log.Debugf("forwarder: Initialization complete with NIC %d", nicID)
 	return f, nil
 }
 
 func (f *Forwarder) InjectIncomingPacket(payload []byte) error {
-	if len(payload) < header.IPv4MinimumSize {
-		return fmt.Errorf("packet too small: %d bytes", len(payload))
+	if len(payload) == 0 {
+		return fmt.Errorf("empty packet")
+	}
+
+	var protoNum tcpip.NetworkProtocolNumber
+	switch payload[0] >> 4 {
+	case 4:
+		if len(payload) < header.IPv4MinimumSize {
+			return fmt.Errorf("IPv4 packet too small: %d bytes", len(payload))
+		}
+		if f.handleICMPDirect(payload) {
+			return nil
+		}
+		protoNum = ipv4.ProtocolNumber
+	case 6:
+		if len(payload) < header.IPv6MinimumSize {
+			return fmt.Errorf("IPv6 packet too small: %d bytes", len(payload))
+		}
+		if f.handleICMPDirect(payload) {
+			return nil
+		}
+		protoNum = ipv6.ProtocolNumber
+	default:
+		return fmt.Errorf("unknown IP version: %d", payload[0]>>4)
 	}
 
 	pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
@@ -144,11 +202,95 @@ func (f *Forwarder) InjectIncomingPacket(payload []byte) error {
 	defer pkt.DecRef()
 
 	if f.endpoint.dispatcher != nil {
-		f.endpoint.dispatcher.DeliverNetworkPacket(ipv4.ProtocolNumber, pkt)
+		f.endpoint.dispatcher.DeliverNetworkPacket(protoNum, pkt)
 	}
 	return nil
 }
 
+// handleICMPDirect intercepts ICMP packets from raw IP payloads before they
+// enter gVisor. It synthesizes the TransportEndpointID and PacketBuffer that
+// the existing handlers expect, then dispatches to handleICMP/handleICMPv6.
+// This bypasses gVisor's network layer which causes duplicate v4 echo replies
+// and auto-replies to all v6 echo requests in promiscuous mode.
+//
+// Unlike gVisor's network layer, this does not validate ICMP checksums or
+// reassemble IP fragments. Fragmented ICMP packets fall through to gVisor.
+func parseICMPv4(payload []byte) (ipHdrLen int, src, dst tcpip.Address, ok bool) {
+	ip := header.IPv4(payload)
+	if ip.Protocol() != uint8(header.ICMPv4ProtocolNumber) {
+		return 0, src, dst, false
+	}
+	if ip.FragmentOffset() != 0 || ip.Flags()&header.IPv4FlagMoreFragments != 0 {
+		return 0, src, dst, false
+	}
+	ipHdrLen = int(ip.HeaderLength())
+	if len(payload)-ipHdrLen < header.ICMPv4MinimumSize {
+		return 0, src, dst, false
+	}
+	return ipHdrLen, ip.SourceAddress(), ip.DestinationAddress(), true
+}
+
+func parseICMPv6(payload []byte) (ipHdrLen int, src, dst tcpip.Address, ok bool) {
+	ip := header.IPv6(payload)
+	if ip.NextHeader() != uint8(header.ICMPv6ProtocolNumber) {
+		return 0, src, dst, false
+	}
+	ipHdrLen = header.IPv6MinimumSize
+	if len(payload)-ipHdrLen < header.ICMPv6MinimumSize {
+		return 0, src, dst, false
+	}
+	return ipHdrLen, ip.SourceAddress(), ip.DestinationAddress(), true
+}
+
+func (f *Forwarder) handleICMPDirect(payload []byte) bool {
+	var (
+		ipHdrLen int
+		srcAddr  tcpip.Address
+		dstAddr  tcpip.Address
+		ok       bool
+	)
+	switch payload[0] >> 4 {
+	case 4:
+		ipHdrLen, srcAddr, dstAddr, ok = parseICMPv4(payload)
+	case 6:
+		ipHdrLen, srcAddr, dstAddr, ok = parseICMPv6(payload)
+	}
+	if !ok {
+		return false
+	}
+
+	// Let gVisor handle ICMP destined for our own addresses natively.
+	// Its network-layer auto-reply is correct and efficient for local traffic.
+	if f.ip.Equal(dstAddr) || f.ipv6.Equal(dstAddr) {
+		return false
+	}
+
+	id := stack.TransportEndpointID{
+		LocalAddress:  dstAddr,
+		RemoteAddress: srcAddr,
+	}
+
+	// Build a PacketBuffer with headers consumed the same way gVisor would.
+	pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
+		Payload: buffer.MakeWithData(payload),
+	})
+	defer pkt.DecRef()
+
+	if _, ok := pkt.NetworkHeader().Consume(ipHdrLen); !ok {
+		return false
+	}
+
+	icmpPayload := payload[ipHdrLen:]
+	if _, ok := pkt.TransportHeader().Consume(len(icmpPayload)); !ok {
+		return false
+	}
+
+	if payload[0]>>4 == 6 {
+		return f.handleICMPv6(id, pkt)
+	}
+	return f.handleICMP(id, pkt)
+}
+
 // Stop gracefully shuts down the forwarder
 func (f *Forwarder) Stop() {
 	f.cancel()
@@ -161,11 +303,14 @@ func (f *Forwarder) Stop() {
 	f.stack.Wait()
 }
 
-func (f *Forwarder) determineDialAddr(addr tcpip.Address) net.IP {
+func (f *Forwarder) determineDialAddr(addr tcpip.Address) netip.Addr {
 	if f.netstack && f.ip.Equal(addr) {
-		return net.IPv4(127, 0, 0, 1)
+		return netip.AddrFrom4([4]byte{127, 0, 0, 1})
 	}
-	return addr.AsSlice()
+	if f.netstack && f.ipv6.Equal(addr) {
+		return netip.IPv6Loopback()
+	}
+	return addrToNetipAddr(addr)
 }
 
 func (f *Forwarder) RegisterRuleID(srcIP, dstIP netip.Addr, srcPort, dstPort uint16, ruleID []byte) {
@@ -198,3 +343,51 @@ func buildKey(srcIP, dstIP netip.Addr, srcPort, dstPort uint16) conntrack.ConnKe
 		DstPort: dstPort,
 	}
 }
+
+// addrFromNetipAddr converts a netip.Addr to a gvisor tcpip.Address without allocating.
+func addrFromNetipAddr(addr netip.Addr) tcpip.Address {
+	if !addr.IsValid() {
+		return tcpip.Address{}
+	}
+	if addr.Is4() {
+		return tcpip.AddrFrom4(addr.As4())
+	}
+	return tcpip.AddrFrom16(addr.As16())
+}
+
+// addrToNetipAddr converts a gvisor tcpip.Address to netip.Addr without allocating.
+func addrToNetipAddr(addr tcpip.Address) netip.Addr {
+	switch addr.Len() {
+	case 4:
+		return netip.AddrFrom4(addr.As4())
+	case 16:
+		return netip.AddrFrom16(addr.As16())
+	default:
+		return netip.Addr{}
+	}
+}
+
+// checkICMPCapability tests whether we have raw ICMP socket access at startup.
+func (f *Forwarder) checkICMPCapability() {
+	f.hasRawICMPAccess = probeRawICMP("ip4:icmp", "0.0.0.0", f.logger)
+	f.hasRawICMPv6Access = probeRawICMP("ip6:ipv6-icmp", "::", f.logger)
+}
+
+func probeRawICMP(network, addr string, logger *nblog.Logger) bool {
+	ctx, cancel := context.WithTimeout(context.Background(), 100*time.Millisecond)
+	defer cancel()
+
+	lc := net.ListenConfig{}
+	conn, err := lc.ListenPacket(ctx, network, addr)
+	if err != nil {
+		logger.Debug1("forwarder: no raw %s socket access, will use ping binary fallback", network)
+		return false
+	}
+
+	if err := conn.Close(); err != nil {
+		logger.Debug2("forwarder: failed to close %s capability test socket: %v", network, err)
+	}
+
+	logger.Debug1("forwarder: raw %s socket access available", network)
+	return true
+}

client/firewall/uspfilter/forwarder/icmp.go

@@ -2,8 +2,11 @@ package forwarder
 
 import (
 	"context"
+	"fmt"
 	"net"
 	"net/netip"
+	"os/exec"
+	"runtime"
 	"time"
 
 	"github.com/google/uuid"
@@ -14,30 +17,100 @@ import (
 )
 
 // handleICMP handles ICMP packets from the network stack
-func (f *Forwarder) handleICMP(id stack.TransportEndpointID, pkt stack.PacketBufferPtr) bool {
+func (f *Forwarder) handleICMP(id stack.TransportEndpointID, pkt *stack.PacketBuffer) bool {
 	icmpHdr := header.ICMPv4(pkt.TransportHeader().View().AsSlice())
-	icmpType := uint8(icmpHdr.Type())
-	icmpCode := uint8(icmpHdr.Code())
-
-	if header.ICMPv4Type(icmpType) == header.ICMPv4EchoReply {
-		// dont process our own replies
-		return true
-	}
 
 	flowID := uuid.New()
-	f.sendICMPEvent(nftypes.TypeStart, flowID, id, icmpType, icmpCode, 0, 0)
+	f.sendICMPEvent(nftypes.TypeStart, flowID, id, uint8(icmpHdr.Type()), uint8(icmpHdr.Code()), 0, 0)
 
-	ctx, cancel := context.WithTimeout(f.ctx, 5*time.Second)
+	// For Echo Requests, send and wait for response
+	if icmpHdr.Type() == header.ICMPv4Echo {
+		return f.handleICMPEcho(flowID, id, pkt, uint8(icmpHdr.Type()), uint8(icmpHdr.Code()))
+	}
+
+	// For other ICMP types (Time Exceeded, Destination Unreachable, etc), forward without waiting
+	if !f.hasRawICMPAccess {
+		f.logger.Debug2("forwarder: Cannot handle ICMP type %v without raw socket access for %v", icmpHdr.Type(), epID(id))
+		return false
+	}
+
+	icmpData := stack.PayloadSince(pkt.TransportHeader()).AsSlice()
+	conn, err := f.forwardICMPPacket(id, icmpData, uint8(icmpHdr.Type()), uint8(icmpHdr.Code()), false, 100*time.Millisecond)
+	if err != nil {
+		f.logger.Error2("forwarder: Failed to forward ICMP packet for %v: %v", epID(id), err)
+		return true
+	}
+	if err := conn.Close(); err != nil {
+		f.logger.Debug1("forwarder: Failed to close ICMP socket: %v", err)
+	}
+
+	return true
+}
+
+// handleICMPEcho handles ICMP echo requests asynchronously with rate limiting.
+func (f *Forwarder) handleICMPEcho(flowID uuid.UUID, id stack.TransportEndpointID, pkt *stack.PacketBuffer, icmpType, icmpCode uint8) bool {
+	select {
+	case f.pingSemaphore <- struct{}{}:
+		icmpData := stack.PayloadSince(pkt.TransportHeader()).ToSlice()
+		rxBytes := pkt.Size()
+
+		go func() {
+			defer func() { <-f.pingSemaphore }()
+
+			if f.hasRawICMPAccess {
+				f.handleICMPViaSocket(flowID, id, icmpType, icmpCode, icmpData, rxBytes)
+			} else {
+				f.handleICMPViaPing(flowID, id, icmpType, icmpCode, icmpData, rxBytes)
+			}
+		}()
+	default:
+		f.logger.Debug3("forwarder: ICMP rate limit exceeded for %v type %v code %v",
+			epID(id), icmpType, icmpCode)
+	}
+	return true
+}
+
+// forwardICMPPacket creates a raw ICMP socket and sends the packet, returning the connection.
+// The caller is responsible for closing the returned connection.
+func (f *Forwarder) forwardICMPPacket(id stack.TransportEndpointID, payload []byte, icmpType, icmpCode uint8, v6 bool, timeout time.Duration) (net.PacketConn, error) {
+	ctx, cancel := context.WithTimeout(f.ctx, timeout)
 	defer cancel()
 
-	lc := net.ListenConfig{}
-	// TODO: support non-root
-	conn, err := lc.ListenPacket(ctx, "ip4:icmp", "0.0.0.0")
-	if err != nil {
-		f.logger.Error2("forwarder: Failed to create ICMP socket for %v: %v", epID(id), err)
+	network, listenAddr := "ip4:icmp", "0.0.0.0"
+	if v6 {
+		network, listenAddr = "ip6:ipv6-icmp", "::"
+	}
 
-		// This will make netstack reply on behalf of the original destination, that's ok for now
-		return false
+	lc := net.ListenConfig{}
+	conn, err := lc.ListenPacket(ctx, network, listenAddr)
+	if err != nil {
+		return nil, fmt.Errorf("create ICMP socket: %w", err)
+	}
+
+	dstIP := f.determineDialAddr(id.LocalAddress)
+	dst := &net.IPAddr{IP: dstIP.AsSlice()}
+
+	if _, err = conn.WriteTo(payload, dst); err != nil {
+		if closeErr := conn.Close(); closeErr != nil {
+			f.logger.Debug1("forwarder: Failed to close ICMP socket: %v", closeErr)
+		}
+		return nil, fmt.Errorf("write ICMP packet: %w", err)
+	}
+
+	f.logger.Trace3("forwarder: Forwarded ICMP packet %v type %v code %v",
+		epID(id), icmpType, icmpCode)
+
+	return conn, nil
+}
+
+// handleICMPViaSocket handles ICMP echo requests using raw sockets.
+func (f *Forwarder) handleICMPViaSocket(flowID uuid.UUID, id stack.TransportEndpointID, icmpType, icmpCode uint8, icmpData []byte, rxBytes int) {
+	sendTime := time.Now()
+
+	conn, err := f.forwardICMPPacket(id, icmpData, icmpType, icmpCode, false, 5*time.Second)
+	if err != nil {
+		f.logger.Error2("forwarder: Failed to send ICMP packet for %v: %v", epID(id), err)
+		return
 	}
 	defer func() {
 		if err := conn.Close(); err != nil {
@@ -45,38 +118,22 @@ func (f *Forwarder) handleICMP(id stack.TransportEndpointID, pkt stack.PacketBuf
 		}
 	}()
 
-	dstIP := f.determineDialAddr(id.LocalAddress)
-	dst := &net.IPAddr{IP: dstIP}
+	txBytes := f.handleEchoResponse(conn, id)
+	rtt := time.Since(sendTime).Round(10 * time.Microsecond)
 
-	fullPacket := stack.PayloadSince(pkt.TransportHeader())
-	payload := fullPacket.AsSlice()
+	f.logger.Trace4("forwarder: Forwarded ICMP echo reply %v type %v code %v (rtt=%v, raw socket)",
+		epID(id), icmpType, icmpCode, rtt)
 
-	if _, err = conn.WriteTo(payload, dst); err != nil {
-		f.logger.Error2("forwarder: Failed to write ICMP packet for %v: %v", epID(id), err)
-		return true
-	}
-
-	f.logger.Trace3("forwarder: Forwarded ICMP packet %v type %v code %v",
-		epID(id), icmpHdr.Type(), icmpHdr.Code())
-
-	// For Echo Requests, send and handle response
-	if header.ICMPv4Type(icmpType) == header.ICMPv4Echo {
-		rxBytes := pkt.Size()
-		txBytes := f.handleEchoResponse(icmpHdr, conn, id)
-		f.sendICMPEvent(nftypes.TypeEnd, flowID, id, icmpType, icmpCode, uint64(rxBytes), uint64(txBytes))
-	}
-
-	// For other ICMP types (Time Exceeded, Destination Unreachable, etc) do nothing
-	return true
+	f.sendICMPEvent(nftypes.TypeEnd, flowID, id, icmpType, icmpCode, uint64(rxBytes), uint64(txBytes))
 }
 
-func (f *Forwarder) handleEchoResponse(icmpHdr header.ICMPv4, conn net.PacketConn, id stack.TransportEndpointID) int {
+func (f *Forwarder) handleEchoResponse(conn net.PacketConn, id stack.TransportEndpointID) int {
 	if err := conn.SetReadDeadline(time.Now().Add(5 * time.Second)); err != nil {
 		f.logger.Error1("forwarder: Failed to set read deadline for ICMP response: %v", err)
 		return 0
 	}
 
-	response := make([]byte, f.endpoint.mtu)
+	response := make([]byte, f.endpoint.mtu.Load())
 	n, _, err := conn.ReadFrom(response)
 	if err != nil {
 		if !isTimeout(err) {
@@ -85,31 +142,7 @@ func (f *Forwarder) handleEchoResponse(icmpHdr header.ICMPv4, conn net.PacketCon
 		return 0
 	}
 
-	ipHdr := make([]byte, header.IPv4MinimumSize)
-	ip := header.IPv4(ipHdr)
-	ip.Encode(&header.IPv4Fields{
-		TotalLength: uint16(header.IPv4MinimumSize + n),
-		TTL:         64,
-		Protocol:    uint8(header.ICMPv4ProtocolNumber),
-		SrcAddr:     id.LocalAddress,
-		DstAddr:     id.RemoteAddress,
-	})
-	ip.SetChecksum(^ip.CalculateChecksum())
-
-	fullPacket := make([]byte, 0, len(ipHdr)+n)
-	fullPacket = append(fullPacket, ipHdr...)
-	fullPacket = append(fullPacket, response[:n]...)
-
-	if err := f.InjectIncomingPacket(fullPacket); err != nil {
-		f.logger.Error1("forwarder: Failed to inject ICMP response: %v", err)
-
-		return 0
-	}
-
-	f.logger.Trace3("forwarder: Forwarded ICMP echo reply for %v type %v code %v",
-		epID(id), icmpHdr.Type(), icmpHdr.Code())
-
-	return len(fullPacket)
+	return f.injectICMPReply(id, response[:n])
 }
 
 // sendICMPEvent stores flow events for ICMP packets
@@ -122,19 +155,23 @@ func (f *Forwarder) sendICMPEvent(typ nftypes.Type, flowID uuid.UUID, id stack.T
 		txPackets = 1
 	}
 
-	srcIp := netip.AddrFrom4(id.RemoteAddress.As4())
-	dstIp := netip.AddrFrom4(id.LocalAddress.As4())
+	srcIp := addrToNetipAddr(id.RemoteAddress)
+	dstIp := addrToNetipAddr(id.LocalAddress)
+
+	proto := nftypes.ICMP
+	if srcIp.Is6() {
+		proto = nftypes.ICMPv6
+	}
 
 	fields := nftypes.EventFields{
 		FlowID:    flowID,
 		Type:      typ,
 		Direction: nftypes.Ingress,
-		Protocol:  nftypes.ICMP,
-		// TODO: handle ipv6
-		SourceIP: srcIp,
-		DestIP:   dstIp,
-		ICMPType: icmpType,
-		ICMPCode: icmpCode,
+		Protocol:  proto,
+		SourceIP:  srcIp,
+		DestIP:    dstIp,
+		ICMPType:  icmpType,
+		ICMPCode:  icmpCode,
 
 		RxBytes:   rxBytes,
 		TxBytes:   txBytes,
@@ -152,3 +189,228 @@ func (f *Forwarder) sendICMPEvent(typ nftypes.Type, flowID uuid.UUID, id stack.T
 
 	f.flowLogger.StoreEvent(fields)
 }
+
+// handleICMPViaPing handles ICMP echo requests by executing the system ping binary.
+// This is used as a fallback when raw socket access is not available.
+func (f *Forwarder) handleICMPViaPing(flowID uuid.UUID, id stack.TransportEndpointID, icmpType, icmpCode uint8, icmpData []byte, rxBytes int) {
+	ctx, cancel := context.WithTimeout(f.ctx, 5*time.Second)
+	defer cancel()
+
+	dstIP := f.determineDialAddr(id.LocalAddress)
+	cmd := buildPingCommand(ctx, dstIP, 5*time.Second)
+
+	pingStart := time.Now()
+	if err := cmd.Run(); err != nil {
+		f.logger.Warn4("forwarder: Ping binary failed for %v type %v code %v: %v", epID(id),
+			icmpType, icmpCode, err)
+		return
+	}
+	rtt := time.Since(pingStart).Round(10 * time.Microsecond)
+
+	f.logger.Trace3("forwarder: Forwarded ICMP echo request %v type %v code %v",
+		epID(id), icmpType, icmpCode)
+
+	txBytes := f.synthesizeEchoReply(id, icmpData)
+
+	f.logger.Trace4("forwarder: Forwarded ICMP echo reply %v type %v code %v (rtt=%v, ping binary)",
+		epID(id), icmpType, icmpCode, rtt)
+
+	f.sendICMPEvent(nftypes.TypeEnd, flowID, id, icmpType, icmpCode, uint64(rxBytes), uint64(txBytes))
+}
+
+// handleICMPv6 handles ICMPv6 packets from the network stack.
+func (f *Forwarder) handleICMPv6(id stack.TransportEndpointID, pkt *stack.PacketBuffer) bool {
+	icmpHdr := header.ICMPv6(pkt.TransportHeader().View().AsSlice())
+
+	flowID := uuid.New()
+	f.sendICMPEvent(nftypes.TypeStart, flowID, id, uint8(icmpHdr.Type()), uint8(icmpHdr.Code()), 0, 0)
+
+	if icmpHdr.Type() == header.ICMPv6EchoRequest {
+		return f.handleICMPv6Echo(flowID, id, pkt, uint8(icmpHdr.Type()), uint8(icmpHdr.Code()))
+	}
+
+	// For non-echo types (Destination Unreachable, Packet Too Big, etc), forward without waiting
+	if !f.hasRawICMPv6Access {
+		f.logger.Debug2("forwarder: Cannot handle ICMPv6 type %v without raw socket access for %v", icmpHdr.Type(), epID(id))
+		return false
+	}
+
+	icmpData := stack.PayloadSince(pkt.TransportHeader()).AsSlice()
+	conn, err := f.forwardICMPPacket(id, icmpData, uint8(icmpHdr.Type()), uint8(icmpHdr.Code()), true, 100*time.Millisecond)
+	if err != nil {
+		f.logger.Error2("forwarder: Failed to forward ICMPv6 packet for %v: %v", epID(id), err)
+		return true
+	}
+	if err := conn.Close(); err != nil {
+		f.logger.Debug1("forwarder: Failed to close ICMPv6 socket: %v", err)
+	}
+
+	return true
+}
+
+// handleICMPv6Echo handles ICMPv6 echo requests using the ping binary.
+func (f *Forwarder) handleICMPv6Echo(flowID uuid.UUID, id stack.TransportEndpointID, pkt *stack.PacketBuffer, icmpType, icmpCode uint8) bool {
+	select {
+	case f.pingSemaphore <- struct{}{}:
+		icmpData := stack.PayloadSince(pkt.TransportHeader()).ToSlice()
+		rxBytes := pkt.Size()
+
+		go func() {
+			defer func() { <-f.pingSemaphore }()
+			f.handleICMPv6ViaPing(flowID, id, icmpType, icmpCode, icmpData, rxBytes)
+		}()
+	default:
+		f.logger.Debug3("forwarder: ICMPv6 rate limit exceeded for %v type %v code %v", epID(id), icmpType, icmpCode)
+	}
+	return true
+}
+
+// handleICMPv6ViaPing uses the system ping6 binary for ICMPv6 echo.
+func (f *Forwarder) handleICMPv6ViaPing(flowID uuid.UUID, id stack.TransportEndpointID, icmpType, icmpCode uint8, icmpData []byte, rxBytes int) {
+	ctx, cancel := context.WithTimeout(f.ctx, 5*time.Second)
+	defer cancel()
+
+	dstIP := f.determineDialAddr(id.LocalAddress)
+	cmd := buildPingCommand(ctx, dstIP, 5*time.Second)
+
+	pingStart := time.Now()
+	if err := cmd.Run(); err != nil {
+		f.logger.Warn4("forwarder: Ping6 failed for %v type %v code %v: %v", epID(id), icmpType, icmpCode, err)
+		return
+	}
+	rtt := time.Since(pingStart).Round(10 * time.Microsecond)
+
+	f.logger.Trace3("forwarder: Forwarded ICMPv6 echo request %v type %v code %v",
+		epID(id), icmpType, icmpCode)
+
+	txBytes := f.synthesizeICMPv6EchoReply(id, icmpData)
+
+	f.logger.Trace4("forwarder: Forwarded ICMPv6 echo reply %v type %v code %v (rtt=%v, ping binary)",
+		epID(id), icmpType, icmpCode, rtt)
+
+	f.sendICMPEvent(nftypes.TypeEnd, flowID, id, icmpType, icmpCode, uint64(rxBytes), uint64(txBytes))
+}
+
+// synthesizeICMPv6EchoReply creates an ICMPv6 echo reply and injects it back.
+func (f *Forwarder) synthesizeICMPv6EchoReply(id stack.TransportEndpointID, icmpData []byte) int {
+	replyICMP := make([]byte, len(icmpData))
+	copy(replyICMP, icmpData)
+
+	replyHdr := header.ICMPv6(replyICMP)
+	replyHdr.SetType(header.ICMPv6EchoReply)
+	replyHdr.SetChecksum(0)
+	// ICMPv6Checksum computes the pseudo-header internally from Src/Dst.
+	// Header contains the full ICMP message, so PayloadCsum/PayloadLen are zero.
+	replyHdr.SetChecksum(header.ICMPv6Checksum(header.ICMPv6ChecksumParams{
+		Header: replyHdr,
+		Src:    id.LocalAddress,
+		Dst:    id.RemoteAddress,
+	}))
+
+	return f.injectICMPv6Reply(id, replyICMP)
+}
+
+// injectICMPv6Reply wraps an ICMPv6 payload in an IPv6 header and sends to the peer.
+func (f *Forwarder) injectICMPv6Reply(id stack.TransportEndpointID, icmpPayload []byte) int {
+	ipHdr := make([]byte, header.IPv6MinimumSize)
+	ip := header.IPv6(ipHdr)
+	ip.Encode(&header.IPv6Fields{
+		PayloadLength:     uint16(len(icmpPayload)),
+		TransportProtocol: header.ICMPv6ProtocolNumber,
+		HopLimit:          64,
+		SrcAddr:           id.LocalAddress,
+		DstAddr:           id.RemoteAddress,
+	})
+
+	fullPacket := make([]byte, 0, len(ipHdr)+len(icmpPayload))
+	fullPacket = append(fullPacket, ipHdr...)
+	fullPacket = append(fullPacket, icmpPayload...)
+
+	if err := f.endpoint.device.CreateOutboundPacket(fullPacket, id.RemoteAddress.AsSlice()); err != nil {
+		f.logger.Error1("forwarder: Failed to send ICMPv6 reply to peer: %v", err)
+		return 0
+	}
+
+	return len(fullPacket)
+}
+
+const (
+	pingBin  = "ping"
+	ping6Bin = "ping6"
+)
+
+// buildPingCommand creates a platform-specific ping command.
+// Most platforms auto-detect IPv6 from raw addresses. macOS/iOS/OpenBSD require ping6.
+func buildPingCommand(ctx context.Context, target netip.Addr, timeout time.Duration) *exec.Cmd {
+	timeoutSec := int(timeout.Seconds())
+	if timeoutSec < 1 {
+		timeoutSec = 1
+	}
+
+	isV6 := target.Is6()
+	timeoutStr := fmt.Sprintf("%d", timeoutSec)
+
+	switch runtime.GOOS {
+	case "linux", "android":
+		return exec.CommandContext(ctx, pingBin, "-c", "1", "-W", timeoutStr, "-q", target.String())
+	case "darwin", "ios":
+		bin := pingBin
+		if isV6 {
+			bin = ping6Bin
+		}
+		return exec.CommandContext(ctx, bin, "-c", "1", "-t", timeoutStr, "-q", target.String())
+	case "freebsd":
+		return exec.CommandContext(ctx, pingBin, "-c", "1", "-t", timeoutStr, target.String())
+	case "openbsd", "netbsd":
+		bin := pingBin
+		if isV6 {
+			bin = ping6Bin
+		}
+		return exec.CommandContext(ctx, bin, "-c", "1", "-w", timeoutStr, target.String())
+	case "windows":
+		return exec.CommandContext(ctx, pingBin, "-n", "1", "-w", fmt.Sprintf("%d", timeoutSec*1000), target.String())
+	default:
+		return exec.CommandContext(ctx, pingBin, "-c", "1", target.String())
+	}
+}
+
+// synthesizeEchoReply creates an ICMP echo reply from raw ICMP data and injects it back into the network stack.
+// Returns the size of the injected packet.
+func (f *Forwarder) synthesizeEchoReply(id stack.TransportEndpointID, icmpData []byte) int {
+	replyICMP := make([]byte, len(icmpData))
+	copy(replyICMP, icmpData)
+
+	replyICMPHdr := header.ICMPv4(replyICMP)
+	replyICMPHdr.SetType(header.ICMPv4EchoReply)
+	replyICMPHdr.SetChecksum(0)
+	replyICMPHdr.SetChecksum(header.ICMPv4Checksum(replyICMPHdr, 0))
+
+	return f.injectICMPReply(id, replyICMP)
+}
+
+// injectICMPReply wraps an ICMP payload in an IP header and injects it into the network stack.
+// Returns the total size of the injected packet, or 0 if injection failed.
+func (f *Forwarder) injectICMPReply(id stack.TransportEndpointID, icmpPayload []byte) int {
+	ipHdr := make([]byte, header.IPv4MinimumSize)
+	ip := header.IPv4(ipHdr)
+	ip.Encode(&header.IPv4Fields{
+		TotalLength: uint16(header.IPv4MinimumSize + len(icmpPayload)),
+		TTL:         64,
+		Protocol:    uint8(header.ICMPv4ProtocolNumber),
+		SrcAddr:     id.LocalAddress,
+		DstAddr:     id.RemoteAddress,
+	})
+	ip.SetChecksum(^ip.CalculateChecksum())
+
+	fullPacket := make([]byte, 0, len(ipHdr)+len(icmpPayload))
+	fullPacket = append(fullPacket, ipHdr...)
+	fullPacket = append(fullPacket, icmpPayload...)
+
+	// Bypass netstack and send directly to peer to avoid looping through our ICMP handler
+	if err := f.endpoint.device.CreateOutboundPacket(fullPacket, id.RemoteAddress.AsSlice()); err != nil {
+		f.logger.Error1("forwarder: Failed to send ICMP reply to peer: %v", err)
+		return 0
+	}
+
+	return len(fullPacket)
+}

client/firewall/uspfilter/forwarder/tcp.go

@@ -2,10 +2,9 @@ package forwarder
 
 import (
 	"context"
-	"fmt"
 	"io"
 	"net"
-	"net/netip"
+	"strconv"
 	"sync"
 
 	"github.com/google/uuid"
@@ -33,7 +32,7 @@ func (f *Forwarder) handleTCP(r *tcp.ForwarderRequest) {
 		}
 	}()
 
-	dialAddr := fmt.Sprintf("%s:%d", f.determineDialAddr(id.LocalAddress), id.LocalPort)
+	dialAddr := net.JoinHostPort(f.determineDialAddr(id.LocalAddress).String(), strconv.Itoa(int(id.LocalPort)))
 
 	outConn, err := (&net.Dialer{}).DialContext(f.ctx, "tcp", dialAddr)
 	if err != nil {
@@ -133,15 +132,14 @@ func (f *Forwarder) proxyTCP(id stack.TransportEndpointID, inConn *gonet.TCPConn
 }
 
 func (f *Forwarder) sendTCPEvent(typ nftypes.Type, flowID uuid.UUID, id stack.TransportEndpointID, rxBytes, txBytes, rxPackets, txPackets uint64) {
-	srcIp := netip.AddrFrom4(id.RemoteAddress.As4())
-	dstIp := netip.AddrFrom4(id.LocalAddress.As4())
+	srcIp := addrToNetipAddr(id.RemoteAddress)
+	dstIp := addrToNetipAddr(id.LocalAddress)
 
 	fields := nftypes.EventFields{
-		FlowID:    flowID,
-		Type:      typ,
-		Direction: nftypes.Ingress,
-		Protocol:  nftypes.TCP,
-		// TODO: handle ipv6
+		FlowID:     flowID,
+		Type:       typ,
+		Direction:  nftypes.Ingress,
+		Protocol:   nftypes.TCP,
 		SourceIP:   srcIp,
 		DestIP:     dstIp,
 		SourcePort: id.RemotePort,

client/firewall/uspfilter/forwarder/udp.go

@@ -4,8 +4,9 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"io"
 	"net"
-	"net/netip"
+	"strconv"
 	"sync"
 	"sync/atomic"
 	"time"
@@ -131,10 +132,10 @@ func (f *udpForwarder) cleanup() {
 }
 
 // handleUDP is called by the UDP forwarder for new packets
-func (f *Forwarder) handleUDP(r *udp.ForwarderRequest) {
+func (f *Forwarder) handleUDP(r *udp.ForwarderRequest) bool {
 	if f.ctx.Err() != nil {
 		f.logger.Trace("forwarder: context done, dropping UDP packet")
-		return
+		return false
 	}
 
 	id := r.ID()
@@ -144,7 +145,7 @@ func (f *Forwarder) handleUDP(r *udp.ForwarderRequest) {
 	f.udpForwarder.RUnlock()
 	if exists {
 		f.logger.Trace1("forwarder: existing UDP connection for %v", epID(id))
-		return
+		return true
 	}
 
 	flowID := uuid.New()
@@ -157,12 +158,12 @@ func (f *Forwarder) handleUDP(r *udp.ForwarderRequest) {
 		}
 	}()
 
-	dstAddr := fmt.Sprintf("%s:%d", f.determineDialAddr(id.LocalAddress), id.LocalPort)
+	dstAddr := net.JoinHostPort(f.determineDialAddr(id.LocalAddress).String(), strconv.Itoa(int(id.LocalPort)))
 	outConn, err := (&net.Dialer{}).DialContext(f.ctx, "udp", dstAddr)
 	if err != nil {
 		f.logger.Debug2("forwarder: UDP dial error for %v: %v", epID(id), err)
 		// TODO: Send ICMP error message
-		return
+		return false
 	}
 
 	// Create wait queue for blocking syscalls
@@ -173,10 +174,10 @@ func (f *Forwarder) handleUDP(r *udp.ForwarderRequest) {
 		if err := outConn.Close(); err != nil {
 			f.logger.Debug2("forwarder: UDP outConn close error for %v: %v", epID(id), err)
 		}
-		return
+		return false
 	}
 
-	inConn := gonet.NewUDPConn(f.stack, &wq, ep)
+	inConn := gonet.NewUDPConn(&wq, ep)
 	connCtx, connCancel := context.WithCancel(f.ctx)
 
 	pConn := &udpPacketConn{
@@ -199,7 +200,7 @@ func (f *Forwarder) handleUDP(r *udp.ForwarderRequest) {
 		if err := outConn.Close(); err != nil {
 			f.logger.Debug2("forwarder: UDP outConn close error for %v: %v", epID(id), err)
 		}
-		return
+		return true
 	}
 	f.udpForwarder.conns[id] = pConn
 	f.udpForwarder.Unlock()
@@ -208,6 +209,7 @@ func (f *Forwarder) handleUDP(r *udp.ForwarderRequest) {
 	f.logger.Trace1("forwarder: established UDP connection %v", epID(id))
 
 	go f.proxyUDP(connCtx, pConn, id, ep)
+	return true
 }
 
 func (f *Forwarder) proxyUDP(ctx context.Context, pConn *udpPacketConn, id stack.TransportEndpointID, ep tcpip.Endpoint) {
@@ -274,15 +276,14 @@ func (f *Forwarder) proxyUDP(ctx context.Context, pConn *udpPacketConn, id stack
 
 // sendUDPEvent stores flow events for UDP connections
 func (f *Forwarder) sendUDPEvent(typ nftypes.Type, flowID uuid.UUID, id stack.TransportEndpointID, rxBytes, txBytes, rxPackets, txPackets uint64) {
-	srcIp := netip.AddrFrom4(id.RemoteAddress.As4())
-	dstIp := netip.AddrFrom4(id.LocalAddress.As4())
+	srcIp := addrToNetipAddr(id.RemoteAddress)
+	dstIp := addrToNetipAddr(id.LocalAddress)
 
 	fields := nftypes.EventFields{
-		FlowID:    flowID,
-		Type:      typ,
-		Direction: nftypes.Ingress,
-		Protocol:  nftypes.UDP,
-		// TODO: handle ipv6
+		FlowID:     flowID,
+		Type:       typ,
+		Direction:  nftypes.Ingress,
+		Protocol:   nftypes.UDP,
 		SourceIP:   srcIp,
 		DestIP:     dstIp,
 		SourcePort: id.RemotePort,
@@ -348,7 +349,7 @@ func (c *udpPacketConn) copy(ctx context.Context, dst net.Conn, src net.Conn, bu
 }
 
 func isClosedError(err error) bool {
-	return errors.Is(err, net.ErrClosed) || errors.Is(err, context.Canceled)
+	return errors.Is(err, net.ErrClosed) || errors.Is(err, context.Canceled) || errors.Is(err, io.EOF)
 }
 
 func isTimeout(err error) bool {

client/firewall/uspfilter/localip.go

@@ -4,89 +4,32 @@ import (
 	"fmt"
 	"net"
 	"net/netip"
-	"sync"
+	"sync/atomic"
 
 	log "github.com/sirupsen/logrus"
 
 	"github.com/netbirdio/netbird/client/firewall/uspfilter/common"
 )
 
-type localIPManager struct {
-	mu sync.RWMutex
-
-	// fixed-size high array for upper byte of a IPv4 address
-	ipv4Bitmap [256]*ipv4LowBitmap
+// localIPSnapshot is an immutable snapshot of local IP addresses, swapped
+// atomically so reads are lock-free.
+type localIPSnapshot struct {
+	ips map[netip.Addr]struct{}
 }
 
-// ipv4LowBitmap is a map for the low 16 bits of a IPv4 address
-type ipv4LowBitmap struct {
-	bitmap [8192]uint32
+type localIPManager struct {
+	snapshot atomic.Pointer[localIPSnapshot]
 }
 
 func newLocalIPManager() *localIPManager {
-	return &localIPManager{}
+	m := &localIPManager{}
+	m.snapshot.Store(&localIPSnapshot{
+		ips: make(map[netip.Addr]struct{}),
+	})
+	return m
 }
 
-func (m *localIPManager) setBitmapBit(ip net.IP) {
-	ipv4 := ip.To4()
-	if ipv4 == nil {
-		return
-	}
-	high := uint16(ipv4[0])
-	low := (uint16(ipv4[1]) << 8) | (uint16(ipv4[2]) << 4) | uint16(ipv4[3])
-
-	index := low / 32
-	bit := low % 32
-
-	if m.ipv4Bitmap[high] == nil {
-		m.ipv4Bitmap[high] = &ipv4LowBitmap{}
-	}
-
-	m.ipv4Bitmap[high].bitmap[index] |= 1 << bit
-}
-
-func (m *localIPManager) setBitInBitmap(ip netip.Addr, bitmap *[256]*ipv4LowBitmap, ipv4Set map[netip.Addr]struct{}, ipv4Addresses *[]netip.Addr) {
-	if !ip.Is4() {
-		return
-	}
-	ipv4 := ip.AsSlice()
-
-	high := uint16(ipv4[0])
-	low := (uint16(ipv4[1]) << 8) | (uint16(ipv4[2]) << 4) | uint16(ipv4[3])
-
-	if bitmap[high] == nil {
-		bitmap[high] = &ipv4LowBitmap{}
-	}
-
-	index := low / 32
-	bit := low % 32
-	bitmap[high].bitmap[index] |= 1 << bit
-
-	if _, exists := ipv4Set[ip]; !exists {
-		ipv4Set[ip] = struct{}{}
-		*ipv4Addresses = append(*ipv4Addresses, ip)
-	}
-}
-
-func (m *localIPManager) checkBitmapBit(ip []byte) bool {
-	high := uint16(ip[0])
-	low := (uint16(ip[1]) << 8) | (uint16(ip[2]) << 4) | uint16(ip[3])
-
-	if m.ipv4Bitmap[high] == nil {
-		return false
-	}
-
-	index := low / 32
-	bit := low % 32
-	return (m.ipv4Bitmap[high].bitmap[index] & (1 << bit)) != 0
-}
-
-func (m *localIPManager) processIP(ip netip.Addr, bitmap *[256]*ipv4LowBitmap, ipv4Set map[netip.Addr]struct{}, ipv4Addresses *[]netip.Addr) error {
-	m.setBitInBitmap(ip, bitmap, ipv4Set, ipv4Addresses)
-	return nil
-}
-
-func (m *localIPManager) processInterface(iface net.Interface, bitmap *[256]*ipv4LowBitmap, ipv4Set map[netip.Addr]struct{}, ipv4Addresses *[]netip.Addr) {
+func processInterface(iface net.Interface, ips map[netip.Addr]struct{}, addresses *[]netip.Addr) {
 	addrs, err := iface.Addrs()
 	if err != nil {
 		log.Debugf("get addresses for interface %s failed: %v", iface.Name, err)
@@ -104,18 +47,19 @@ func (m *localIPManager) processInterface(iface net.Interface, bitmap *[256]*ipv
 			continue
 		}
 
-		addr, ok := netip.AddrFromSlice(ip)
+		parsed, ok := netip.AddrFromSlice(ip)
 		if !ok {
 			log.Warnf("invalid IP address %s in interface %s", ip.String(), iface.Name)
 			continue
 		}
 
-		if err := m.processIP(addr.Unmap(), bitmap, ipv4Set, ipv4Addresses); err != nil {
-			log.Debugf("process IP failed: %v", err)
-		}
+		parsed = parsed.Unmap()
+		ips[parsed] = struct{}{}
+		*addresses = append(*addresses, parsed)
 	}
 }
 
+// UpdateLocalIPs rebuilds the local IP snapshot and swaps it in atomically.
 func (m *localIPManager) UpdateLocalIPs(iface common.IFaceMapper) (err error) {
 	defer func() {
 		if r := recover(); r != nil {
@@ -123,19 +67,20 @@ func (m *localIPManager) UpdateLocalIPs(iface common.IFaceMapper) (err error) {
 		}
 	}()
 
-	var newIPv4Bitmap [256]*ipv4LowBitmap
-	ipv4Set := make(map[netip.Addr]struct{})
-	var ipv4Addresses []netip.Addr
+	ips := make(map[netip.Addr]struct{})
+	var addresses []netip.Addr
 
-	// 127.0.0.0/8
-	newIPv4Bitmap[127] = &ipv4LowBitmap{}
-	for i := 0; i < 8192; i++ {
-		newIPv4Bitmap[127].bitmap[i] = 0xFFFFFFFF
-	}
+	// loopback
+	ips[netip.AddrFrom4([4]byte{127, 0, 0, 1})] = struct{}{}
+	ips[netip.IPv6Loopback()] = struct{}{}
 
 	if iface != nil {
-		if err := m.processIP(iface.Address().IP, &newIPv4Bitmap, ipv4Set, &ipv4Addresses); err != nil {
-			return err
+		ip := iface.Address().IP
+		ips[ip] = struct{}{}
+		addresses = append(addresses, ip)
+		if v6 := iface.Address().IPv6; v6.IsValid() {
+			ips[v6] = struct{}{}
+			addresses = append(addresses, v6)
 		}
 	}
 
@@ -144,25 +89,24 @@ func (m *localIPManager) UpdateLocalIPs(iface common.IFaceMapper) (err error) {
 		log.Warnf("failed to get interfaces: %v", err)
 	} else {
 		for _, intf := range interfaces {
-			m.processInterface(intf, &newIPv4Bitmap, ipv4Set, &ipv4Addresses)
+			processInterface(intf, ips, &addresses)
 		}
 	}
 
-	m.mu.Lock()
-	m.ipv4Bitmap = newIPv4Bitmap
-	m.mu.Unlock()
+	m.snapshot.Store(&localIPSnapshot{ips: ips})
 
-	log.Debugf("Local IPv4 addresses: %v", ipv4Addresses)
+	log.Debugf("Local IP addresses: %v", addresses)
 	return nil
 }
 
+// IsLocalIP checks if the given IP is a local address. Lock-free on the read path.
 func (m *localIPManager) IsLocalIP(ip netip.Addr) bool {
-	if !ip.Is4() {
-		return false
+	s := m.snapshot.Load()
+
+	if ip.Is4() && ip.As4()[0] == 127 {
+		return true
 	}
 
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-
-	return m.checkBitmapBit(ip.AsSlice())
+	_, found := s.ips[ip]
+	return found
 }

client/firewall/uspfilter/localip_bench_test.go

@@ -0,0 +1,72 @@
+package uspfilter
+
+import (
+	"net/netip"
+	"testing"
+
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+)
+
+func setupManager(b *testing.B) *localIPManager {
+	b.Helper()
+	m := newLocalIPManager()
+	mock := &IFaceMock{
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("100.64.0.1"),
+				Network: netip.MustParsePrefix("100.64.0.0/16"),
+				IPv6:    netip.MustParseAddr("fd00::1"),
+				IPv6Net: netip.MustParsePrefix("fd00::/64"),
+			}
+		},
+	}
+	if err := m.UpdateLocalIPs(mock); err != nil {
+		b.Fatalf("UpdateLocalIPs: %v", err)
+	}
+	return m
+}
+
+func BenchmarkIsLocalIP_v4_hit(b *testing.B) {
+	m := setupManager(b)
+	ip := netip.MustParseAddr("100.64.0.1")
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		m.IsLocalIP(ip)
+	}
+}
+
+func BenchmarkIsLocalIP_v4_miss(b *testing.B) {
+	m := setupManager(b)
+	ip := netip.MustParseAddr("8.8.8.8")
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		m.IsLocalIP(ip)
+	}
+}
+
+func BenchmarkIsLocalIP_v6_hit(b *testing.B) {
+	m := setupManager(b)
+	ip := netip.MustParseAddr("fd00::1")
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		m.IsLocalIP(ip)
+	}
+}
+
+func BenchmarkIsLocalIP_v6_miss(b *testing.B) {
+	m := setupManager(b)
+	ip := netip.MustParseAddr("2001:db8::1")
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		m.IsLocalIP(ip)
+	}
+}
+
+func BenchmarkIsLocalIP_loopback(b *testing.B) {
+	m := setupManager(b)
+	ip := netip.MustParseAddr("127.0.0.1")
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		m.IsLocalIP(ip)
+	}
+}

client/firewall/uspfilter/localip_test.go

@@ -72,14 +72,45 @@ func TestLocalIPManager(t *testing.T) {
 			expected: false,
 		},
 		{
-			name: "IPv6 address",
+			name: "IPv6 address matches",
 			setupAddr: wgaddr.Address{
-				IP:      netip.MustParseAddr("fe80::1"),
+				IP:      netip.MustParseAddr("100.64.0.1"),
+				Network: netip.MustParsePrefix("100.64.0.0/16"),
+				IPv6:    netip.MustParseAddr("fd00::1"),
+				IPv6Net: netip.MustParsePrefix("fd00::/64"),
+			},
+			testIP:   netip.MustParseAddr("fd00::1"),
+			expected: true,
+		},
+		{
+			name: "IPv6 address does not match",
+			setupAddr: wgaddr.Address{
+				IP:      netip.MustParseAddr("100.64.0.1"),
+				Network: netip.MustParsePrefix("100.64.0.0/16"),
+				IPv6:    netip.MustParseAddr("fd00::1"),
+				IPv6Net: netip.MustParsePrefix("fd00::/64"),
+			},
+			testIP:   netip.MustParseAddr("fd00::99"),
+			expected: false,
+		},
+		{
+			name: "No aliasing between similar IPs",
+			setupAddr: wgaddr.Address{
+				IP:      netip.MustParseAddr("192.168.1.1"),
 				Network: netip.MustParsePrefix("192.168.1.0/24"),
 			},
-			testIP:   netip.MustParseAddr("fe80::1"),
+			testIP:   netip.MustParseAddr("192.168.0.17"),
 			expected: false,
 		},
+		{
+			name: "IPv6 loopback",
+			setupAddr: wgaddr.Address{
+				IP:      netip.MustParseAddr("100.64.0.1"),
+				Network: netip.MustParsePrefix("100.64.0.0/16"),
+			},
+			testIP:   netip.MustParseAddr("::1"),
+			expected: true,
+		},
 	}
 
 	for _, tt := range tests {
@@ -171,90 +202,3 @@ func TestLocalIPManager_AllInterfaces(t *testing.T) {
 		})
 	}
 }
-
-// MapImplementation is a version using map[string]struct{}
-type MapImplementation struct {
-	localIPs map[string]struct{}
-}
-
-func BenchmarkIPChecks(b *testing.B) {
-	interfaces := make([]net.IP, 16)
-	for i := range interfaces {
-		interfaces[i] = net.IPv4(10, 0, byte(i>>8), byte(i))
-	}
-
-	// Setup bitmap
-	bitmapManager := newLocalIPManager()
-	for _, ip := range interfaces[:8] { // Add half of IPs
-		bitmapManager.setBitmapBit(ip)
-	}
-
-	// Setup map version
-	mapManager := &MapImplementation{
-		localIPs: make(map[string]struct{}),
-	}
-	for _, ip := range interfaces[:8] {
-		mapManager.localIPs[ip.String()] = struct{}{}
-	}
-
-	b.Run("Bitmap_Hit", func(b *testing.B) {
-		ip := interfaces[4]
-		b.ResetTimer()
-		for i := 0; i < b.N; i++ {
-			bitmapManager.checkBitmapBit(ip)
-		}
-	})
-
-	b.Run("Bitmap_Miss", func(b *testing.B) {
-		ip := interfaces[12]
-		b.ResetTimer()
-		for i := 0; i < b.N; i++ {
-			bitmapManager.checkBitmapBit(ip)
-		}
-	})
-
-	b.Run("Map_Hit", func(b *testing.B) {
-		ip := interfaces[4]
-		b.ResetTimer()
-		for i := 0; i < b.N; i++ {
-			// nolint:gosimple
-			_, _ = mapManager.localIPs[ip.String()]
-		}
-	})
-
-	b.Run("Map_Miss", func(b *testing.B) {
-		ip := interfaces[12]
-		b.ResetTimer()
-		for i := 0; i < b.N; i++ {
-			// nolint:gosimple
-			_, _ = mapManager.localIPs[ip.String()]
-		}
-	})
-}
-
-func BenchmarkWGPosition(b *testing.B) {
-	wgIP := net.ParseIP("10.10.0.1")
-
-	// Create two managers - one checks WG IP first, other checks it last
-	b.Run("WG_First", func(b *testing.B) {
-		bm := newLocalIPManager()
-		bm.setBitmapBit(wgIP)
-		b.ResetTimer()
-		for i := 0; i < b.N; i++ {
-			bm.checkBitmapBit(wgIP)
-		}
-	})
-
-	b.Run("WG_Last", func(b *testing.B) {
-		bm := newLocalIPManager()
-		// Fill with other IPs first
-		for i := 0; i < 15; i++ {
-			bm.setBitmapBit(net.IPv4(10, 0, byte(i>>8), byte(i)))
-		}
-		bm.setBitmapBit(wgIP) // Add WG IP last
-		b.ResetTimer()
-		for i := 0; i < b.N; i++ {
-			bm.checkBitmapBit(wgIP)
-		}
-	})
-}

client/firewall/uspfilter/log/log.go

@@ -5,6 +5,8 @@ import (
 	"context"
 	"fmt"
 	"io"
+	"os"
+	"strconv"
 	"sync"
 	"sync/atomic"
 	"time"
@@ -16,9 +18,18 @@ const (
 	maxBatchSize         = 1024 * 16
 	maxMessageSize       = 1024 * 2
 	defaultFlushInterval = 2 * time.Second
-	logChannelSize       = 1000
+	defaultLogChanSize   = 1000
 )
 
+func getLogChannelSize() int {
+	if v := os.Getenv("NB_USPFILTER_LOG_BUFFER"); v != "" {
+		if n, err := strconv.Atoi(v); err == nil && n > 0 {
+			return n
+		}
+	}
+	return defaultLogChanSize
+}
+
 type Level uint32
 
 const (
@@ -69,7 +80,7 @@ type Logger struct {
 func NewFromLogrus(logrusLogger *log.Logger) *Logger {
 	l := &Logger{
 		output:     logrusLogger.Out,
-		msgChannel: make(chan logMessage, logChannelSize),
+		msgChannel: make(chan logMessage, getLogChannelSize()),
 		shutdown:   make(chan struct{}),
 		bufPool: sync.Pool{
 			New: func() any {
@@ -168,6 +179,15 @@ func (l *Logger) Warn3(format string, arg1, arg2, arg3 any) {
 	}
 }
 
+func (l *Logger) Warn4(format string, arg1, arg2, arg3, arg4 any) {
+	if l.level.Load() >= uint32(LevelWarn) {
+		select {
+		case l.msgChannel <- logMessage{level: LevelWarn, format: format, arg1: arg1, arg2: arg2, arg3: arg3, arg4: arg4}:
+		default:
+		}
+	}
+}
+
 func (l *Logger) Debug1(format string, arg1 any) {
 	if l.level.Load() >= uint32(LevelDebug) {
 		select {

client/firewall/uspfilter/nat.go

@@ -13,8 +13,6 @@ import (
 	firewall "github.com/netbirdio/netbird/client/firewall/manager"
 )
 
-var ErrIPv4Only = errors.New("only IPv4 is supported for DNAT")
-
 var (
 	errInvalidIPHeaderLength = errors.New("invalid IP header length")
 )
@@ -25,10 +23,33 @@ const (
 	destinationPortOffset = 2
 
 	// IP address offsets in IPv4 header
-	sourceIPOffset      = 12
-	destinationIPOffset = 16
+	ipv4SrcOffset = 12
+	ipv4DstOffset = 16
+
+	// IP address offsets in IPv6 header
+	ipv6SrcOffset = 8
+	ipv6DstOffset = 24
+
+	// IPv6 fixed header length
+	ipv6HeaderLen = 40
 )
 
+// ipHeaderLen returns the IP header length based on the decoded layer type.
+func ipHeaderLen(d *decoder) (int, error) {
+	switch d.decoded[0] {
+	case layers.LayerTypeIPv4:
+		n := int(d.ip4.IHL) * 4
+		if n < 20 {
+			return 0, errInvalidIPHeaderLength
+		}
+		return n, nil
+	case layers.LayerTypeIPv6:
+		return ipv6HeaderLen, nil
+	default:
+		return 0, fmt.Errorf("unknown IP layer: %v", d.decoded[0])
+	}
+}
+
 // ipv4Checksum calculates IPv4 header checksum.
 func ipv4Checksum(header []byte) uint16 {
 	if len(header) < 20 {
@@ -234,14 +255,13 @@ func (m *Manager) translateOutboundDNAT(packetData []byte, d *decoder) bool {
 		return false
 	}
 
-	dstIP := netip.AddrFrom4([4]byte{packetData[16], packetData[17], packetData[18], packetData[19]})
-
+	_, dstIP := extractPacketIPs(packetData, d)
 	translatedIP, exists := m.getDNATTranslation(dstIP)
 	if !exists {
 		return false
 	}
 
-	if err := m.rewritePacketIP(packetData, d, translatedIP, destinationIPOffset); err != nil {
+	if err := m.rewritePacketIP(packetData, d, translatedIP, false); err != nil {
 		m.logger.Error1("failed to rewrite packet destination: %v", err)
 		return false
 	}
@@ -256,14 +276,13 @@ func (m *Manager) translateInboundReverse(packetData []byte, d *decoder) bool {
 		return false
 	}
 
-	srcIP := netip.AddrFrom4([4]byte{packetData[12], packetData[13], packetData[14], packetData[15]})
-
+	srcIP, _ := extractPacketIPs(packetData, d)
 	originalIP, exists := m.findReverseDNATMapping(srcIP)
 	if !exists {
 		return false
 	}
 
-	if err := m.rewritePacketIP(packetData, d, originalIP, sourceIPOffset); err != nil {
+	if err := m.rewritePacketIP(packetData, d, originalIP, true); err != nil {
 		m.logger.Error1("failed to rewrite packet source: %v", err)
 		return false
 	}
@@ -272,38 +291,96 @@ func (m *Manager) translateInboundReverse(packetData []byte, d *decoder) bool {
 	return true
 }
 
-// rewritePacketIP replaces an IP address (source or destination) in the packet and updates checksums.
-func (m *Manager) rewritePacketIP(packetData []byte, d *decoder, newIP netip.Addr, ipOffset int) error {
+// extractPacketIPs extracts src and dst IP addresses directly from raw packet bytes.
+func extractPacketIPs(packetData []byte, d *decoder) (src, dst netip.Addr) {
+	switch d.decoded[0] {
+	case layers.LayerTypeIPv4:
+		src = netip.AddrFrom4([4]byte{packetData[ipv4SrcOffset], packetData[ipv4SrcOffset+1], packetData[ipv4SrcOffset+2], packetData[ipv4SrcOffset+3]})
+		dst = netip.AddrFrom4([4]byte{packetData[ipv4DstOffset], packetData[ipv4DstOffset+1], packetData[ipv4DstOffset+2], packetData[ipv4DstOffset+3]})
+	case layers.LayerTypeIPv6:
+		src = netip.AddrFrom16([16]byte(packetData[ipv6SrcOffset : ipv6SrcOffset+16]))
+		dst = netip.AddrFrom16([16]byte(packetData[ipv6DstOffset : ipv6DstOffset+16]))
+	}
+	return src, dst
+}
+
+// rewritePacketIP replaces a source (isSource=true) or destination IP address in the packet and updates checksums.
+func (m *Manager) rewritePacketIP(packetData []byte, d *decoder, newIP netip.Addr, isSource bool) error {
+	hdrLen, err := ipHeaderLen(d)
+	if err != nil {
+		return err
+	}
+
+	switch d.decoded[0] {
+	case layers.LayerTypeIPv4:
+		return m.rewriteIPv4(packetData, d, newIP, hdrLen, isSource)
+	case layers.LayerTypeIPv6:
+		return m.rewriteIPv6(packetData, d, newIP, hdrLen, isSource)
+	default:
+		return fmt.Errorf("unknown IP layer: %v", d.decoded[0])
+	}
+}
+
+func (m *Manager) rewriteIPv4(packetData []byte, d *decoder, newIP netip.Addr, hdrLen int, isSource bool) error {
 	if !newIP.Is4() {
-		return ErrIPv4Only
+		return fmt.Errorf("cannot write IPv6 address into IPv4 packet")
+	}
+
+	offset := ipv4DstOffset
+	if isSource {
+		offset = ipv4SrcOffset
 	}
 
 	var oldIP [4]byte
-	copy(oldIP[:], packetData[ipOffset:ipOffset+4])
+	copy(oldIP[:], packetData[offset:offset+4])
 	newIPBytes := newIP.As4()
+	copy(packetData[offset:offset+4], newIPBytes[:])
 
-	copy(packetData[ipOffset:ipOffset+4], newIPBytes[:])
-
-	ipHeaderLen := int(d.ip4.IHL) * 4
-	if ipHeaderLen < 20 || ipHeaderLen > len(packetData) {
-		return errInvalidIPHeaderLength
-	}
-
+	// Recalculate IPv4 header checksum
 	binary.BigEndian.PutUint16(packetData[10:12], 0)
-	ipChecksum := ipv4Checksum(packetData[:ipHeaderLen])
-	binary.BigEndian.PutUint16(packetData[10:12], ipChecksum)
+	binary.BigEndian.PutUint16(packetData[10:12], ipv4Checksum(packetData[:hdrLen]))
 
+	// Update transport checksums incrementally
 	if len(d.decoded) > 1 {
 		switch d.decoded[1] {
 		case layers.LayerTypeTCP:
-			m.updateTCPChecksum(packetData, ipHeaderLen, oldIP[:], newIPBytes[:])
+			m.updateTCPChecksum(packetData, hdrLen, oldIP[:], newIPBytes[:])
 		case layers.LayerTypeUDP:
-			m.updateUDPChecksum(packetData, ipHeaderLen, oldIP[:], newIPBytes[:])
+			m.updateUDPChecksum(packetData, hdrLen, oldIP[:], newIPBytes[:])
 		case layers.LayerTypeICMPv4:
-			m.updateICMPChecksum(packetData, ipHeaderLen)
+			m.updateICMPChecksum(packetData, hdrLen)
 		}
 	}
+	return nil
+}
 
+func (m *Manager) rewriteIPv6(packetData []byte, d *decoder, newIP netip.Addr, hdrLen int, isSource bool) error {
+	if !newIP.Is6() {
+		return fmt.Errorf("cannot write IPv4 address into IPv6 packet")
+	}
+
+	offset := ipv6DstOffset
+	if isSource {
+		offset = ipv6SrcOffset
+	}
+
+	var oldIP [16]byte
+	copy(oldIP[:], packetData[offset:offset+16])
+	newIPBytes := newIP.As16()
+	copy(packetData[offset:offset+16], newIPBytes[:])
+
+	// IPv6 has no header checksum, only update transport checksums
+	if len(d.decoded) > 1 {
+		switch d.decoded[1] {
+		case layers.LayerTypeTCP:
+			m.updateTCPChecksum(packetData, hdrLen, oldIP[:], newIPBytes[:])
+		case layers.LayerTypeUDP:
+			m.updateUDPChecksum(packetData, hdrLen, oldIP[:], newIPBytes[:])
+		case layers.LayerTypeICMPv6:
+			// ICMPv6 checksum includes pseudo-header with addresses, use incremental update
+			m.updateICMPv6Checksum(packetData, hdrLen, oldIP[:], newIPBytes[:])
+		}
+	}
 	return nil
 }
 
@@ -351,6 +428,20 @@ func (m *Manager) updateICMPChecksum(packetData []byte, ipHeaderLen int) {
 	binary.BigEndian.PutUint16(icmpData[2:4], checksum)
 }
 
+// updateICMPv6Checksum updates ICMPv6 checksum after address change.
+// ICMPv6 uses a pseudo-header (like TCP/UDP), so incremental update applies.
+func (m *Manager) updateICMPv6Checksum(packetData []byte, ipHeaderLen int, oldIP, newIP []byte) {
+	icmpStart := ipHeaderLen
+	if len(packetData) < icmpStart+4 {
+		return
+	}
+
+	checksumOffset := icmpStart + 2
+	oldChecksum := binary.BigEndian.Uint16(packetData[checksumOffset : checksumOffset+2])
+	newChecksum := incrementalUpdate(oldChecksum, oldIP, newIP)
+	binary.BigEndian.PutUint16(packetData[checksumOffset:checksumOffset+2], newChecksum)
+}
+
 // incrementalUpdate performs incremental checksum update per RFC 1624.
 func incrementalUpdate(oldChecksum uint16, oldBytes, newBytes []byte) uint16 {
 	sum := uint32(^oldChecksum)
@@ -358,9 +449,9 @@ func incrementalUpdate(oldChecksum uint16, oldBytes, newBytes []byte) uint16 {
 	// Fast path for IPv4 addresses (4 bytes) - most common case
 	if len(oldBytes) == 4 && len(newBytes) == 4 {
 		sum += uint32(^binary.BigEndian.Uint16(oldBytes[0:2]))
-		sum += uint32(^binary.BigEndian.Uint16(oldBytes[2:4]))
+		sum += uint32(^binary.BigEndian.Uint16(oldBytes[2:4])) //nolint:gosec // length checked above
 		sum += uint32(binary.BigEndian.Uint16(newBytes[0:2]))
-		sum += uint32(binary.BigEndian.Uint16(newBytes[2:4]))
+		sum += uint32(binary.BigEndian.Uint16(newBytes[2:4])) //nolint:gosec // length checked above
 	} else {
 		// Fallback for other lengths
 		for i := 0; i < len(oldBytes)-1; i += 2 {
@@ -515,12 +606,12 @@ func (m *Manager) applyPortRule(packetData []byte, d *decoder, srcIP, dstIP neti
 
 // rewriteTCPPort rewrites a TCP port (source or destination) and updates checksum.
 func (m *Manager) rewriteTCPPort(packetData []byte, d *decoder, newPort uint16, portOffset int) error {
-	ipHeaderLen := int(d.ip4.IHL) * 4
-	if ipHeaderLen < 20 || ipHeaderLen > len(packetData) {
-		return errInvalidIPHeaderLength
+	hdrLen, err := ipHeaderLen(d)
+	if err != nil {
+		return err
 	}
 
-	tcpStart := ipHeaderLen
+	tcpStart := hdrLen
 	if len(packetData) < tcpStart+4 {
 		return fmt.Errorf("packet too short for TCP header")
 	}
@@ -546,12 +637,12 @@ func (m *Manager) rewriteTCPPort(packetData []byte, d *decoder, newPort uint16,
 
 // rewriteUDPPort rewrites a UDP port (source or destination) and updates checksum.
 func (m *Manager) rewriteUDPPort(packetData []byte, d *decoder, newPort uint16, portOffset int) error {
-	ipHeaderLen := int(d.ip4.IHL) * 4
-	if ipHeaderLen < 20 || ipHeaderLen > len(packetData) {
-		return errInvalidIPHeaderLength
+	hdrLen, err := ipHeaderLen(d)
+	if err != nil {
+		return err
 	}
 
-	udpStart := ipHeaderLen
+	udpStart := hdrLen
 	if len(packetData) < udpStart+8 {
 		return fmt.Errorf("packet too short for UDP header")
 	}

client/firewall/uspfilter/nat_bench_test.go

@@ -342,12 +342,17 @@ func BenchmarkDNATMemoryAllocations(b *testing.B) {
 
 		// Parse the packet fresh each time to get a clean decoder
 		d := &decoder{decoded: []gopacket.LayerType{}}
-		d.parser = gopacket.NewDecodingLayerParser(
+		d.parser4 = gopacket.NewDecodingLayerParser(
 			layers.LayerTypeIPv4,
 			&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
 		)
-		d.parser.IgnoreUnsupported = true
-		err = d.parser.DecodeLayers(testPacket, &d.decoded)
+		d.parser4.IgnoreUnsupported = true
+		d.parser6 = gopacket.NewDecodingLayerParser(
+			layers.LayerTypeIPv6,
+			&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
+		)
+		d.parser6.IgnoreUnsupported = true
+		err = d.decodePacket(testPacket)
 		assert.NoError(b, err)
 
 		manager.translateOutboundDNAT(testPacket, d)
@@ -371,12 +376,17 @@ func BenchmarkDirectIPExtraction(b *testing.B) {
 	b.Run("decoder_extraction", func(b *testing.B) {
 		// Create decoder once for comparison
 		d := &decoder{decoded: []gopacket.LayerType{}}
-		d.parser = gopacket.NewDecodingLayerParser(
+		d.parser4 = gopacket.NewDecodingLayerParser(
 			layers.LayerTypeIPv4,
 			&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
 		)
-		d.parser.IgnoreUnsupported = true
-		err := d.parser.DecodeLayers(packet, &d.decoded)
+		d.parser4.IgnoreUnsupported = true
+		d.parser6 = gopacket.NewDecodingLayerParser(
+			layers.LayerTypeIPv6,
+			&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
+		)
+		d.parser6.IgnoreUnsupported = true
+		err := d.decodePacket(packet)
 		assert.NoError(b, err)
 
 		for i := 0; i < b.N; i++ {

client/firewall/uspfilter/nat_test.go

@@ -86,13 +86,18 @@ func parsePacket(t testing.TB, packetData []byte) *decoder {
 	d := &decoder{
 		decoded: []gopacket.LayerType{},
 	}
-	d.parser = gopacket.NewDecodingLayerParser(
+	d.parser4 = gopacket.NewDecodingLayerParser(
 		layers.LayerTypeIPv4,
 		&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
 	)
-	d.parser.IgnoreUnsupported = true
+	d.parser4.IgnoreUnsupported = true
+	d.parser6 = gopacket.NewDecodingLayerParser(
+		layers.LayerTypeIPv6,
+		&d.eth, &d.ip4, &d.ip6, &d.icmp4, &d.icmp6, &d.tcp, &d.udp,
+	)
+	d.parser6.IgnoreUnsupported = true
 
-	err := d.parser.DecodeLayers(packetData, &d.decoded)
+	err := d.decodePacket(packetData)
 	require.NoError(t, err)
 	return d
 }
@@ -234,9 +239,10 @@ func TestInboundPortDNATNegative(t *testing.T) {
 			require.False(t, translated, "Packet should NOT be translated for %s", tc.name)
 
 			d = parsePacket(t, packet)
-			if tc.protocol == layers.IPProtocolTCP {
+			switch tc.protocol {
+			case layers.IPProtocolTCP:
 				require.Equal(t, tc.dstPort, uint16(d.tcp.DstPort), "Port should remain unchanged")
-			} else if tc.protocol == layers.IPProtocolUDP {
+			case layers.IPProtocolUDP:
 				require.Equal(t, tc.dstPort, uint16(d.udp.DstPort), "Port should remain unchanged")
 			}
 		})

client/firewall/uspfilter/rule.go

@@ -34,7 +34,7 @@ type RouteRule struct {
 	sources      []netip.Prefix
 	dstSet       firewall.Set
 	destinations []netip.Prefix
-	proto        firewall.Protocol
+	protoLayer   gopacket.LayerType
 	srcPort      *firewall.Port
 	dstPort      *firewall.Port
 	action       firewall.Action

client/firewall/uspfilter/tracer.go

@@ -112,10 +112,13 @@ func (t *PacketTrace) AddResultWithForwarder(stage PacketStage, message string,
 }
 
 func (p *PacketBuilder) Build() ([]byte, error) {
-	ip := p.buildIPLayer()
-	pktLayers := []gopacket.SerializableLayer{ip}
+	ipLayer, err := p.buildIPLayer()
+	if err != nil {
+		return nil, err
+	}
+	pktLayers := []gopacket.SerializableLayer{ipLayer}
 
-	transportLayer, err := p.buildTransportLayer(ip)
+	transportLayer, err := p.buildTransportLayer(ipLayer)
 	if err != nil {
 		return nil, err
 	}
@@ -129,30 +132,43 @@ func (p *PacketBuilder) Build() ([]byte, error) {
 	return serializePacket(pktLayers)
 }
 
-func (p *PacketBuilder) buildIPLayer() *layers.IPv4 {
+func (p *PacketBuilder) buildIPLayer() (gopacket.SerializableLayer, error) {
+	if p.SrcIP.Is4() != p.DstIP.Is4() {
+		return nil, fmt.Errorf("mixed address families: src=%s dst=%s", p.SrcIP, p.DstIP)
+	}
+	proto := getIPProtocolNumber(p.Protocol, p.SrcIP.Is6())
+	if p.SrcIP.Is6() {
+		return &layers.IPv6{
+			Version:    6,
+			HopLimit:   64,
+			NextHeader: proto,
+			SrcIP:      p.SrcIP.AsSlice(),
+			DstIP:      p.DstIP.AsSlice(),
+		}, nil
+	}
 	return &layers.IPv4{
 		Version:  4,
 		TTL:      64,
-		Protocol: layers.IPProtocol(getIPProtocolNumber(p.Protocol)),
+		Protocol: proto,
 		SrcIP:    p.SrcIP.AsSlice(),
 		DstIP:    p.DstIP.AsSlice(),
-	}
+	}, nil
 }
 
-func (p *PacketBuilder) buildTransportLayer(ip *layers.IPv4) ([]gopacket.SerializableLayer, error) {
+func (p *PacketBuilder) buildTransportLayer(ipLayer gopacket.SerializableLayer) ([]gopacket.SerializableLayer, error) {
 	switch p.Protocol {
 	case "tcp":
-		return p.buildTCPLayer(ip)
+		return p.buildTCPLayer(ipLayer)
 	case "udp":
-		return p.buildUDPLayer(ip)
+		return p.buildUDPLayer(ipLayer)
 	case "icmp":
-		return p.buildICMPLayer()
+		return p.buildICMPLayer(ipLayer)
 	default:
 		return nil, fmt.Errorf("unsupported protocol: %s", p.Protocol)
 	}
 }
 
-func (p *PacketBuilder) buildTCPLayer(ip *layers.IPv4) ([]gopacket.SerializableLayer, error) {
+func (p *PacketBuilder) buildTCPLayer(ipLayer gopacket.SerializableLayer) ([]gopacket.SerializableLayer, error) {
 	tcp := &layers.TCP{
 		SrcPort: layers.TCPPort(p.SrcPort),
 		DstPort: layers.TCPPort(p.DstPort),
@@ -164,24 +180,37 @@ func (p *PacketBuilder) buildTCPLayer(ip *layers.IPv4) ([]gopacket.SerializableL
 		PSH:     p.TCPState != nil && p.TCPState.PSH,
 		URG:     p.TCPState != nil && p.TCPState.URG,
 	}
-	if err := tcp.SetNetworkLayerForChecksum(ip); err != nil {
-		return nil, fmt.Errorf("set network layer for TCP checksum: %w", err)
+	if nl, ok := ipLayer.(gopacket.NetworkLayer); ok {
+		if err := tcp.SetNetworkLayerForChecksum(nl); err != nil {
+			return nil, fmt.Errorf("set network layer for TCP checksum: %w", err)
+		}
 	}
 	return []gopacket.SerializableLayer{tcp}, nil
 }
 
-func (p *PacketBuilder) buildUDPLayer(ip *layers.IPv4) ([]gopacket.SerializableLayer, error) {
+func (p *PacketBuilder) buildUDPLayer(ipLayer gopacket.SerializableLayer) ([]gopacket.SerializableLayer, error) {
 	udp := &layers.UDP{
 		SrcPort: layers.UDPPort(p.SrcPort),
 		DstPort: layers.UDPPort(p.DstPort),
 	}
-	if err := udp.SetNetworkLayerForChecksum(ip); err != nil {
-		return nil, fmt.Errorf("set network layer for UDP checksum: %w", err)
+	if nl, ok := ipLayer.(gopacket.NetworkLayer); ok {
+		if err := udp.SetNetworkLayerForChecksum(nl); err != nil {
+			return nil, fmt.Errorf("set network layer for UDP checksum: %w", err)
+		}
 	}
 	return []gopacket.SerializableLayer{udp}, nil
 }
 
-func (p *PacketBuilder) buildICMPLayer() ([]gopacket.SerializableLayer, error) {
+func (p *PacketBuilder) buildICMPLayer(ipLayer gopacket.SerializableLayer) ([]gopacket.SerializableLayer, error) {
+	if p.SrcIP.Is6() || p.DstIP.Is6() {
+		icmp := &layers.ICMPv6{
+			TypeCode: layers.CreateICMPv6TypeCode(p.ICMPType, p.ICMPCode),
+		}
+		if nl, ok := ipLayer.(gopacket.NetworkLayer); ok {
+			_ = icmp.SetNetworkLayerForChecksum(nl)
+		}
+		return []gopacket.SerializableLayer{icmp}, nil
+	}
 	icmp := &layers.ICMPv4{
 		TypeCode: layers.CreateICMPv4TypeCode(p.ICMPType, p.ICMPCode),
 	}
@@ -204,14 +233,17 @@ func serializePacket(layers []gopacket.SerializableLayer) ([]byte, error) {
 	return buf.Bytes(), nil
 }
 
-func getIPProtocolNumber(protocol fw.Protocol) int {
+func getIPProtocolNumber(protocol fw.Protocol, isV6 bool) layers.IPProtocol {
 	switch protocol {
 	case fw.ProtocolTCP:
-		return int(layers.IPProtocolTCP)
+		return layers.IPProtocolTCP
 	case fw.ProtocolUDP:
-		return int(layers.IPProtocolUDP)
+		return layers.IPProtocolUDP
 	case fw.ProtocolICMP:
-		return int(layers.IPProtocolICMPv4)
+		if isV6 {
+			return layers.IPProtocolICMPv6
+		}
+		return layers.IPProtocolICMPv4
 	default:
 		return 0
 	}
@@ -234,7 +266,7 @@ func (m *Manager) TracePacket(packetData []byte, direction fw.RuleDirection) *Pa
 	trace := &PacketTrace{Direction: direction}
 
 	// Initial packet decoding
-	if err := d.parser.DecodeLayers(packetData, &d.decoded); err != nil {
+	if err := d.decodePacket(packetData); err != nil {
 		trace.AddResult(StageReceived, fmt.Sprintf("Failed to decode packet: %v", err), false)
 		return trace
 	}
@@ -256,6 +288,8 @@ func (m *Manager) TracePacket(packetData []byte, direction fw.RuleDirection) *Pa
 		trace.DestinationPort = uint16(d.udp.DstPort)
 	case layers.LayerTypeICMPv4:
 		trace.Protocol = "ICMP"
+	case layers.LayerTypeICMPv6:
+		trace.Protocol = "ICMPv6"
 	}
 
 	trace.AddResult(StageReceived, fmt.Sprintf("Received %s packet: %s:%d -> %s:%d",
@@ -379,9 +413,9 @@ func (m *Manager) handleNativeRouter(trace *PacketTrace) *PacketTrace {
 }
 
 func (m *Manager) handleRouteACLs(trace *PacketTrace, d *decoder, srcIP, dstIP netip.Addr) *PacketTrace {
-	proto, _ := getProtocolFromPacket(d)
+	protoLayer := d.decoded[1]
 	srcPort, dstPort := getPortsFromPacket(d)
-	id, allowed := m.routeACLsPass(srcIP, dstIP, proto, srcPort, dstPort)
+	id, allowed := m.routeACLsPass(srcIP, dstIP, protoLayer, srcPort, dstPort)
 
 	strId := string(id)
 	if id == nil {
@@ -415,7 +449,7 @@ func (m *Manager) traceOutbound(packetData []byte, trace *PacketTrace) *PacketTr
 	d := m.decoders.Get().(*decoder)
 	defer m.decoders.Put(d)
 
-	if err := d.parser.DecodeLayers(packetData, &d.decoded); err != nil {
+	if err := d.decodePacket(packetData); err != nil {
 		trace.AddResult(StageCompleted, "Packet dropped - decode error", false)
 		return trace
 	}
@@ -434,7 +468,7 @@ func (m *Manager) traceOutbound(packetData []byte, trace *PacketTrace) *PacketTr
 func (m *Manager) handleInboundDNAT(trace *PacketTrace, packetData []byte, d *decoder, srcIP, dstIP *netip.Addr) bool {
 	portDNATApplied := m.traceInboundPortDNAT(trace, packetData, d)
 	if portDNATApplied {
-		if err := d.parser.DecodeLayers(packetData, &d.decoded); err != nil {
+		if err := d.decodePacket(packetData); err != nil {
 			trace.AddResult(StageInboundPortDNAT, "Failed to re-decode after port DNAT", false)
 			return true
 		}
@@ -444,7 +478,7 @@ func (m *Manager) handleInboundDNAT(trace *PacketTrace, packetData []byte, d *de
 
 	nat1to1Applied := m.traceInbound1to1NAT(trace, packetData, d)
 	if nat1to1Applied {
-		if err := d.parser.DecodeLayers(packetData, &d.decoded); err != nil {
+		if err := d.decodePacket(packetData); err != nil {
 			trace.AddResult(StageInbound1to1NAT, "Failed to re-decode after 1:1 NAT", false)
 			return true
 		}
@@ -509,7 +543,7 @@ func (m *Manager) traceInbound1to1NAT(trace *PacketTrace, packetData []byte, d *
 		return false
 	}
 
-	srcIP := netip.AddrFrom4([4]byte{packetData[12], packetData[13], packetData[14], packetData[15]})
+	srcIP, _ := extractPacketIPs(packetData, d)
 
 	translated := m.translateInboundReverse(packetData, d)
 	if translated {
@@ -539,7 +573,7 @@ func (m *Manager) traceOutbound1to1NAT(trace *PacketTrace, packetData []byte, d
 		return false
 	}
 
-	dstIP := netip.AddrFrom4([4]byte{packetData[16], packetData[17], packetData[18], packetData[19]})
+	_, dstIP := extractPacketIPs(packetData, d)
 
 	translated := m.translateOutboundDNAT(packetData, d)
 	if translated {

client/grpc/dialer.go

@@ -28,7 +28,7 @@ func Backoff(ctx context.Context) backoff.BackOff {
 
 // CreateConnection creates a gRPC client connection with the appropriate transport options.
 // The component parameter specifies the WebSocket proxy component path (e.g., "/management", "/signal").
-func CreateConnection(ctx context.Context, addr string, tlsEnabled bool, component string) (*grpc.ClientConn, error) {
+func CreateConnection(ctx context.Context, addr string, tlsEnabled bool, component string, extraOpts ...grpc.DialOption) (*grpc.ClientConn, error) {
 	transportOption := grpc.WithTransportCredentials(insecure.NewCredentials())
 	// for js, the outer websocket layer takes care of tls
 	if tlsEnabled && runtime.GOOS != "js" {
@@ -46,9 +46,7 @@ func CreateConnection(ctx context.Context, addr string, tlsEnabled bool, compone
 	connCtx, cancel := context.WithTimeout(ctx, 30*time.Second)
 	defer cancel()
 
-	conn, err := grpc.DialContext(
-		connCtx,
-		addr,
+	opts := []grpc.DialOption{
 		transportOption,
 		WithCustomDialer(tlsEnabled, component),
 		grpc.WithBlock(),
@@ -56,7 +54,10 @@ func CreateConnection(ctx context.Context, addr string, tlsEnabled bool, compone
 			Time:    30 * time.Second,
 			Timeout: 10 * time.Second,
 		}),
-	)
+	}
+	opts = append(opts, extraOpts...)
+
+	conn, err := grpc.DialContext(connCtx, addr, opts...)
 	if err != nil {
 		return nil, fmt.Errorf("dial context: %w", err)
 	}

client/iface/bind/dual_stack_conn.go

@@ -0,0 +1,169 @@
+package bind
+
+import (
+	"errors"
+	"net"
+	"sync"
+	"time"
+
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+)
+
+var (
+	errNoIPv4Conn  = errors.New("no IPv4 connection available")
+	errNoIPv6Conn  = errors.New("no IPv6 connection available")
+	errInvalidAddr = errors.New("invalid address type")
+)
+
+// DualStackPacketConn wraps IPv4 and IPv6 UDP connections and routes writes
+// to the appropriate connection based on the destination address.
+// ReadFrom is not used in the hot path - ICEBind receives packets via
+// BatchReader.ReadBatch() directly. This is only used by udpMux for sending.
+type DualStackPacketConn struct {
+	ipv4Conn net.PacketConn
+	ipv6Conn net.PacketConn
+
+	readFromWarn sync.Once
+}
+
+// NewDualStackPacketConn creates a new dual-stack packet connection.
+func NewDualStackPacketConn(ipv4Conn, ipv6Conn net.PacketConn) *DualStackPacketConn {
+	return &DualStackPacketConn{
+		ipv4Conn: ipv4Conn,
+		ipv6Conn: ipv6Conn,
+	}
+}
+
+// ReadFrom reads from the available connection (preferring IPv4).
+// NOTE: This method is NOT used in the data path. ICEBind receives packets via
+// BatchReader.ReadBatch() directly for both IPv4 and IPv6, which is much more efficient.
+// This implementation exists only to satisfy the net.PacketConn interface for the udpMux,
+// but the udpMux only uses WriteTo() for sending STUN responses - it never calls ReadFrom()
+// because STUN packets are filtered and forwarded via HandleSTUNMessage() from the receive path.
+func (d *DualStackPacketConn) ReadFrom(b []byte) (n int, addr net.Addr, err error) {
+	d.readFromWarn.Do(func() {
+		log.Warn("DualStackPacketConn.ReadFrom called - this is unexpected and may indicate an inefficient code path")
+	})
+
+	if d.ipv4Conn != nil {
+		return d.ipv4Conn.ReadFrom(b)
+	}
+	if d.ipv6Conn != nil {
+		return d.ipv6Conn.ReadFrom(b)
+	}
+	return 0, nil, net.ErrClosed
+}
+
+// WriteTo writes to the appropriate connection based on the address type.
+func (d *DualStackPacketConn) WriteTo(b []byte, addr net.Addr) (n int, err error) {
+	udpAddr, ok := addr.(*net.UDPAddr)
+	if !ok {
+		return 0, &net.OpError{
+			Op:   "write",
+			Net:  "udp",
+			Addr: addr,
+			Err:  errInvalidAddr,
+		}
+	}
+
+	if udpAddr.IP.To4() == nil {
+		if d.ipv6Conn != nil {
+			return d.ipv6Conn.WriteTo(b, addr)
+		}
+		return 0, &net.OpError{
+			Op:   "write",
+			Net:  "udp6",
+			Addr: addr,
+			Err:  errNoIPv6Conn,
+		}
+	}
+
+	if d.ipv4Conn != nil {
+		return d.ipv4Conn.WriteTo(b, addr)
+	}
+	return 0, &net.OpError{
+		Op:   "write",
+		Net:  "udp4",
+		Addr: addr,
+		Err:  errNoIPv4Conn,
+	}
+}
+
+// Close closes both connections.
+func (d *DualStackPacketConn) Close() error {
+	var result *multierror.Error
+	if d.ipv4Conn != nil {
+		if err := d.ipv4Conn.Close(); err != nil {
+			result = multierror.Append(result, err)
+		}
+	}
+	if d.ipv6Conn != nil {
+		if err := d.ipv6Conn.Close(); err != nil {
+			result = multierror.Append(result, err)
+		}
+	}
+	return nberrors.FormatErrorOrNil(result)
+}
+
+// LocalAddr returns the local address of the IPv4 connection if available,
+// otherwise the IPv6 connection.
+func (d *DualStackPacketConn) LocalAddr() net.Addr {
+	if d.ipv4Conn != nil {
+		return d.ipv4Conn.LocalAddr()
+	}
+	if d.ipv6Conn != nil {
+		return d.ipv6Conn.LocalAddr()
+	}
+	return nil
+}
+
+// SetDeadline sets the deadline for both connections.
+func (d *DualStackPacketConn) SetDeadline(t time.Time) error {
+	var result *multierror.Error
+	if d.ipv4Conn != nil {
+		if err := d.ipv4Conn.SetDeadline(t); err != nil {
+			result = multierror.Append(result, err)
+		}
+	}
+	if d.ipv6Conn != nil {
+		if err := d.ipv6Conn.SetDeadline(t); err != nil {
+			result = multierror.Append(result, err)
+		}
+	}
+	return nberrors.FormatErrorOrNil(result)
+}
+
+// SetReadDeadline sets the read deadline for both connections.
+func (d *DualStackPacketConn) SetReadDeadline(t time.Time) error {
+	var result *multierror.Error
+	if d.ipv4Conn != nil {
+		if err := d.ipv4Conn.SetReadDeadline(t); err != nil {
+			result = multierror.Append(result, err)
+		}
+	}
+	if d.ipv6Conn != nil {
+		if err := d.ipv6Conn.SetReadDeadline(t); err != nil {
+			result = multierror.Append(result, err)
+		}
+	}
+	return nberrors.FormatErrorOrNil(result)
+}
+
+// SetWriteDeadline sets the write deadline for both connections.
+func (d *DualStackPacketConn) SetWriteDeadline(t time.Time) error {
+	var result *multierror.Error
+	if d.ipv4Conn != nil {
+		if err := d.ipv4Conn.SetWriteDeadline(t); err != nil {
+			result = multierror.Append(result, err)
+		}
+	}
+	if d.ipv6Conn != nil {
+		if err := d.ipv6Conn.SetWriteDeadline(t); err != nil {
+			result = multierror.Append(result, err)
+		}
+	}
+	return nberrors.FormatErrorOrNil(result)
+}

client/iface/bind/dual_stack_conn_bench_test.go

@@ -0,0 +1,119 @@
+package bind
+
+import (
+	"net"
+	"testing"
+)
+
+var (
+	ipv4Addr = &net.UDPAddr{IP: net.ParseIP("127.0.0.1"), Port: 12345}
+	ipv6Addr = &net.UDPAddr{IP: net.ParseIP("::1"), Port: 12345}
+	payload  = make([]byte, 1200)
+)
+
+func BenchmarkWriteTo_DirectUDPConn(b *testing.B) {
+	conn, err := net.ListenUDP("udp4", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		b.Fatal(err)
+	}
+	defer conn.Close()
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_, _ = conn.WriteTo(payload, ipv4Addr)
+	}
+}
+
+func BenchmarkWriteTo_DualStack_IPv4Only(b *testing.B) {
+	conn, err := net.ListenUDP("udp4", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		b.Fatal(err)
+	}
+	defer conn.Close()
+
+	ds := NewDualStackPacketConn(conn, nil)
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_, _ = ds.WriteTo(payload, ipv4Addr)
+	}
+}
+
+func BenchmarkWriteTo_DualStack_IPv6Only(b *testing.B) {
+	conn, err := net.ListenUDP("udp6", &net.UDPAddr{IP: net.IPv6zero, Port: 0})
+	if err != nil {
+		b.Skipf("IPv6 not available: %v", err)
+	}
+	defer conn.Close()
+
+	ds := NewDualStackPacketConn(nil, conn)
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_, _ = ds.WriteTo(payload, ipv6Addr)
+	}
+}
+
+func BenchmarkWriteTo_DualStack_Both_IPv4Traffic(b *testing.B) {
+	conn4, err := net.ListenUDP("udp4", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		b.Fatal(err)
+	}
+	defer conn4.Close()
+
+	conn6, err := net.ListenUDP("udp6", &net.UDPAddr{IP: net.IPv6zero, Port: 0})
+	if err != nil {
+		b.Skipf("IPv6 not available: %v", err)
+	}
+	defer conn6.Close()
+
+	ds := NewDualStackPacketConn(conn4, conn6)
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_, _ = ds.WriteTo(payload, ipv4Addr)
+	}
+}
+
+func BenchmarkWriteTo_DualStack_Both_IPv6Traffic(b *testing.B) {
+	conn4, err := net.ListenUDP("udp4", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		b.Fatal(err)
+	}
+	defer conn4.Close()
+
+	conn6, err := net.ListenUDP("udp6", &net.UDPAddr{IP: net.IPv6zero, Port: 0})
+	if err != nil {
+		b.Skipf("IPv6 not available: %v", err)
+	}
+	defer conn6.Close()
+
+	ds := NewDualStackPacketConn(conn4, conn6)
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_, _ = ds.WriteTo(payload, ipv6Addr)
+	}
+}
+
+func BenchmarkWriteTo_DualStack_Both_MixedTraffic(b *testing.B) {
+	conn4, err := net.ListenUDP("udp4", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		b.Fatal(err)
+	}
+	defer conn4.Close()
+
+	conn6, err := net.ListenUDP("udp6", &net.UDPAddr{IP: net.IPv6zero, Port: 0})
+	if err != nil {
+		b.Skipf("IPv6 not available: %v", err)
+	}
+	defer conn6.Close()
+
+	ds := NewDualStackPacketConn(conn4, conn6)
+	addrs := []net.Addr{ipv4Addr, ipv6Addr}
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_, _ = ds.WriteTo(payload, addrs[i&1])
+	}
+}

client/iface/bind/dual_stack_conn_test.go

@@ -0,0 +1,191 @@
+package bind
+
+import (
+	"net"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestDualStackPacketConn_RoutesWritesToCorrectSocket(t *testing.T) {
+	ipv4Conn := &mockPacketConn{network: "udp4"}
+	ipv6Conn := &mockPacketConn{network: "udp6"}
+	dualStack := NewDualStackPacketConn(ipv4Conn, ipv6Conn)
+
+	tests := []struct {
+		name       string
+		addr       *net.UDPAddr
+		wantSocket string
+	}{
+		{
+			name:       "IPv4 address",
+			addr:       &net.UDPAddr{IP: net.ParseIP("192.168.1.1"), Port: 1234},
+			wantSocket: "udp4",
+		},
+		{
+			name:       "IPv6 address",
+			addr:       &net.UDPAddr{IP: net.ParseIP("2001:db8::1"), Port: 1234},
+			wantSocket: "udp6",
+		},
+		{
+			name:       "IPv4-mapped IPv6 goes to IPv4",
+			addr:       &net.UDPAddr{IP: net.ParseIP("::ffff:192.168.1.1"), Port: 1234},
+			wantSocket: "udp4",
+		},
+		{
+			name:       "IPv4 loopback",
+			addr:       &net.UDPAddr{IP: net.ParseIP("127.0.0.1"), Port: 1234},
+			wantSocket: "udp4",
+		},
+		{
+			name:       "IPv6 loopback",
+			addr:       &net.UDPAddr{IP: net.ParseIP("::1"), Port: 1234},
+			wantSocket: "udp6",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			ipv4Conn.writeCount = 0
+			ipv6Conn.writeCount = 0
+
+			n, err := dualStack.WriteTo([]byte("test"), tt.addr)
+			require.NoError(t, err)
+			assert.Equal(t, 4, n)
+
+			if tt.wantSocket == "udp4" {
+				assert.Equal(t, 1, ipv4Conn.writeCount, "expected write to IPv4")
+				assert.Equal(t, 0, ipv6Conn.writeCount, "expected no write to IPv6")
+			} else {
+				assert.Equal(t, 0, ipv4Conn.writeCount, "expected no write to IPv4")
+				assert.Equal(t, 1, ipv6Conn.writeCount, "expected write to IPv6")
+			}
+		})
+	}
+}
+
+func TestDualStackPacketConn_IPv4OnlyRejectsIPv6(t *testing.T) {
+	dualStack := NewDualStackPacketConn(&mockPacketConn{network: "udp4"}, nil)
+
+	// IPv4 works
+	_, err := dualStack.WriteTo([]byte("test"), &net.UDPAddr{IP: net.ParseIP("192.168.1.1"), Port: 1234})
+	require.NoError(t, err)
+
+	// IPv6 fails
+	_, err = dualStack.WriteTo([]byte("test"), &net.UDPAddr{IP: net.ParseIP("2001:db8::1"), Port: 1234})
+	require.Error(t, err)
+	assert.Contains(t, err.Error(), "no IPv6 connection")
+}
+
+func TestDualStackPacketConn_IPv6OnlyRejectsIPv4(t *testing.T) {
+	dualStack := NewDualStackPacketConn(nil, &mockPacketConn{network: "udp6"})
+
+	// IPv6 works
+	_, err := dualStack.WriteTo([]byte("test"), &net.UDPAddr{IP: net.ParseIP("2001:db8::1"), Port: 1234})
+	require.NoError(t, err)
+
+	// IPv4 fails
+	_, err = dualStack.WriteTo([]byte("test"), &net.UDPAddr{IP: net.ParseIP("192.168.1.1"), Port: 1234})
+	require.Error(t, err)
+	assert.Contains(t, err.Error(), "no IPv4 connection")
+}
+
+// TestDualStackPacketConn_ReadFromIsNotUsedInHotPath documents that ReadFrom
+// only reads from one socket (IPv4 preferred). This is fine because the actual
+// receive path uses wireguard-go's BatchReader directly, not ReadFrom.
+func TestDualStackPacketConn_ReadFromIsNotUsedInHotPath(t *testing.T) {
+	ipv4Conn := &mockPacketConn{
+		network:  "udp4",
+		readData: []byte("from ipv4"),
+		readAddr: &net.UDPAddr{IP: net.ParseIP("192.168.1.1"), Port: 1234},
+	}
+	ipv6Conn := &mockPacketConn{
+		network:  "udp6",
+		readData: []byte("from ipv6"),
+		readAddr: &net.UDPAddr{IP: net.ParseIP("2001:db8::1"), Port: 1234},
+	}
+
+	dualStack := NewDualStackPacketConn(ipv4Conn, ipv6Conn)
+
+	buf := make([]byte, 100)
+	n, addr, err := dualStack.ReadFrom(buf)
+
+	require.NoError(t, err)
+	// reads from IPv4 (preferred) - this is expected behavior
+	assert.Equal(t, "from ipv4", string(buf[:n]))
+	assert.Equal(t, "192.168.1.1", addr.(*net.UDPAddr).IP.String())
+}
+
+func TestDualStackPacketConn_LocalAddrPrefersIPv4(t *testing.T) {
+	ipv4Addr := &net.UDPAddr{IP: net.ParseIP("0.0.0.0"), Port: 51820}
+	ipv6Addr := &net.UDPAddr{IP: net.ParseIP("::"), Port: 51820}
+
+	tests := []struct {
+		name     string
+		ipv4     net.PacketConn
+		ipv6     net.PacketConn
+		wantAddr net.Addr
+	}{
+		{
+			name:     "both available returns IPv4",
+			ipv4:     &mockPacketConn{localAddr: ipv4Addr},
+			ipv6:     &mockPacketConn{localAddr: ipv6Addr},
+			wantAddr: ipv4Addr,
+		},
+		{
+			name:     "IPv4 only",
+			ipv4:     &mockPacketConn{localAddr: ipv4Addr},
+			ipv6:     nil,
+			wantAddr: ipv4Addr,
+		},
+		{
+			name:     "IPv6 only",
+			ipv4:     nil,
+			ipv6:     &mockPacketConn{localAddr: ipv6Addr},
+			wantAddr: ipv6Addr,
+		},
+		{
+			name:     "neither returns nil",
+			ipv4:     nil,
+			ipv6:     nil,
+			wantAddr: nil,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			dualStack := NewDualStackPacketConn(tt.ipv4, tt.ipv6)
+			assert.Equal(t, tt.wantAddr, dualStack.LocalAddr())
+		})
+	}
+}
+
+// mock
+
+type mockPacketConn struct {
+	network    string
+	writeCount int
+	readData   []byte
+	readAddr   net.Addr
+	localAddr  net.Addr
+}
+
+func (m *mockPacketConn) ReadFrom(b []byte) (n int, addr net.Addr, err error) {
+	if m.readData != nil {
+		return copy(b, m.readData), m.readAddr, nil
+	}
+	return 0, nil, nil
+}
+
+func (m *mockPacketConn) WriteTo(b []byte, addr net.Addr) (n int, err error) {
+	m.writeCount++
+	return len(b), nil
+}
+
+func (m *mockPacketConn) Close() error                       { return nil }
+func (m *mockPacketConn) LocalAddr() net.Addr                { return m.localAddr }
+func (m *mockPacketConn) SetDeadline(t time.Time) error      { return nil }
+func (m *mockPacketConn) SetReadDeadline(t time.Time) error  { return nil }
+func (m *mockPacketConn) SetWriteDeadline(t time.Time) error { return nil }

client/iface/bind/ice_bind.go

@@ -14,7 +14,6 @@ import (
 	"github.com/pion/stun/v3"
 	"github.com/pion/transport/v3"
 	log "github.com/sirupsen/logrus"
-	"golang.org/x/net/ipv4"
 	"golang.org/x/net/ipv6"
 	wgConn "golang.zx2c4.com/wireguard/conn"
 
@@ -27,8 +26,8 @@ type receiverCreator struct {
 	iceBind *ICEBind
 }
 
-func (rc receiverCreator) CreateIPv4ReceiverFn(pc *ipv4.PacketConn, conn *net.UDPConn, rxOffload bool, msgPool *sync.Pool) wgConn.ReceiveFunc {
-	return rc.iceBind.createIPv4ReceiverFn(pc, conn, rxOffload, msgPool)
+func (rc receiverCreator) CreateReceiverFn(pc wgConn.BatchReader, conn *net.UDPConn, rxOffload bool, msgPool *sync.Pool) wgConn.ReceiveFunc {
+	return rc.iceBind.createReceiverFn(pc, conn, rxOffload, msgPool)
 }
 
 // ICEBind is a bind implementation with two main features:
@@ -58,6 +57,8 @@ type ICEBind struct {
 
 	muUDPMux sync.Mutex
 	udpMux   *udpmux.UniversalUDPMuxDefault
+	ipv4Conn *net.UDPConn
+	ipv6Conn *net.UDPConn
 }
 
 func NewICEBind(transportNet transport.Net, filterFn udpmux.FilterFn, address wgaddr.Address, mtu uint16) *ICEBind {
@@ -103,6 +104,12 @@ func (s *ICEBind) Close() error {
 
 	close(s.closedChan)
 
+	s.muUDPMux.Lock()
+	s.ipv4Conn = nil
+	s.ipv6Conn = nil
+	s.udpMux = nil
+	s.muUDPMux.Unlock()
+
 	return s.StdNetBind.Close()
 }
 
@@ -160,19 +167,18 @@ func (b *ICEBind) Send(bufs [][]byte, ep wgConn.Endpoint) error {
 	return nil
 }
 
-func (s *ICEBind) createIPv4ReceiverFn(pc *ipv4.PacketConn, conn *net.UDPConn, rxOffload bool, msgsPool *sync.Pool) wgConn.ReceiveFunc {
+func (s *ICEBind) createReceiverFn(pc wgConn.BatchReader, conn *net.UDPConn, rxOffload bool, msgsPool *sync.Pool) wgConn.ReceiveFunc {
 	s.muUDPMux.Lock()
 	defer s.muUDPMux.Unlock()
 
-	s.udpMux = udpmux.NewUniversalUDPMuxDefault(
-		udpmux.UniversalUDPMuxParams{
-			UDPConn:   nbnet.WrapPacketConn(conn),
-			Net:       s.transportNet,
-			FilterFn:  s.filterFn,
-			WGAddress: s.address,
-			MTU:       s.mtu,
-		},
-	)
+	// Detect IPv4 vs IPv6 from connection's local address
+	if localAddr := conn.LocalAddr().(*net.UDPAddr); localAddr.IP.To4() != nil {
+		s.ipv4Conn = conn
+	} else {
+		s.ipv6Conn = conn
+	}
+	s.createOrUpdateMux()
+
 	return func(bufs [][]byte, sizes []int, eps []wgConn.Endpoint) (n int, err error) {
 		msgs := getMessages(msgsPool)
 		for i := range bufs {
@@ -180,12 +186,13 @@ func (s *ICEBind) createIPv4ReceiverFn(pc *ipv4.PacketConn, conn *net.UDPConn, r
 			(*msgs)[i].OOB = (*msgs)[i].OOB[:cap((*msgs)[i].OOB)]
 		}
 		defer putMessages(msgs, msgsPool)
+
 		var numMsgs int
 		if runtime.GOOS == "linux" || runtime.GOOS == "android" {
 			if rxOffload {
 				readAt := len(*msgs) - (wgConn.IdealBatchSize / wgConn.UdpSegmentMaxDatagrams)
-				//nolint
-				numMsgs, err = pc.ReadBatch((*msgs)[readAt:], 0)
+				//nolint:staticcheck
+				_, err = pc.ReadBatch((*msgs)[readAt:], 0)
 				if err != nil {
 					return 0, err
 				}
@@ -207,12 +214,12 @@ func (s *ICEBind) createIPv4ReceiverFn(pc *ipv4.PacketConn, conn *net.UDPConn, r
 			}
 			numMsgs = 1
 		}
+
 		for i := 0; i < numMsgs; i++ {
 			msg := &(*msgs)[i]
 
 			// todo: handle err
-			ok, _ := s.filterOutStunMessages(msg.Buffers, msg.N, msg.Addr)
-			if ok {
+			if ok, _ := s.filterOutStunMessages(msg.Buffers, msg.N, msg.Addr); ok {
 				continue
 			}
 			sizes[i] = msg.N
@@ -233,6 +240,38 @@ func (s *ICEBind) createIPv4ReceiverFn(pc *ipv4.PacketConn, conn *net.UDPConn, r
 	}
 }
 
+// createOrUpdateMux creates or updates the UDP mux with the available connections.
+// Must be called with muUDPMux held.
+func (s *ICEBind) createOrUpdateMux() {
+	var muxConn net.PacketConn
+
+	switch {
+	case s.ipv4Conn != nil && s.ipv6Conn != nil:
+		muxConn = NewDualStackPacketConn(
+			nbnet.WrapPacketConn(s.ipv4Conn),
+			nbnet.WrapPacketConn(s.ipv6Conn),
+		)
+	case s.ipv4Conn != nil:
+		muxConn = nbnet.WrapPacketConn(s.ipv4Conn)
+	case s.ipv6Conn != nil:
+		muxConn = nbnet.WrapPacketConn(s.ipv6Conn)
+	default:
+		return
+	}
+
+	// Don't close the old mux - it doesn't own the underlying connections.
+	// The sockets are managed by WireGuard's StdNetBind, not by us.
+	s.udpMux = udpmux.NewUniversalUDPMuxDefault(
+		udpmux.UniversalUDPMuxParams{
+			UDPConn:   muxConn,
+			Net:       s.transportNet,
+			FilterFn:  s.filterFn,
+			WGAddress: s.address,
+			MTU:       s.mtu,
+		},
+	)
+}
+
 func (s *ICEBind) filterOutStunMessages(buffers [][]byte, n int, addr net.Addr) (bool, error) {
 	for i := range buffers {
 		if !stun.IsMessage(buffers[i]) {
@@ -245,9 +284,14 @@ func (s *ICEBind) filterOutStunMessages(buffers [][]byte, n int, addr net.Addr)
 			return true, err
 		}
 
-		muxErr := s.udpMux.HandleSTUNMessage(msg, addr)
-		if muxErr != nil {
-			log.Warnf("failed to handle STUN packet")
+		s.muUDPMux.Lock()
+		mux := s.udpMux
+		s.muUDPMux.Unlock()
+
+		if mux != nil {
+			if muxErr := mux.HandleSTUNMessage(msg, addr); muxErr != nil {
+				log.Warnf("failed to handle STUN packet: %v", muxErr)
+			}
 		}
 
 		buffers[i] = []byte{}

client/iface/bind/ice_bind_test.go

@@ -0,0 +1,324 @@
+package bind
+
+import (
+	"fmt"
+	"net"
+	"net/netip"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/pion/transport/v3/stdnet"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.org/x/net/ipv4"
+	"golang.org/x/net/ipv6"
+
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+)
+
+func TestICEBind_CreatesReceiverForBothIPv4AndIPv6(t *testing.T) {
+	iceBind := setupICEBind(t)
+
+	ipv4Conn, ipv6Conn := createDualStackConns(t)
+	defer ipv4Conn.Close()
+	defer ipv6Conn.Close()
+
+	rc := receiverCreator{iceBind}
+	pool := createMsgPool()
+
+	// Simulate wireguard-go calling CreateReceiverFn for IPv4
+	ipv4RecvFn := rc.CreateReceiverFn(ipv4.NewPacketConn(ipv4Conn), ipv4Conn, false, pool)
+	require.NotNil(t, ipv4RecvFn)
+
+	iceBind.muUDPMux.Lock()
+	assert.NotNil(t, iceBind.ipv4Conn, "should store IPv4 connection")
+	assert.Nil(t, iceBind.ipv6Conn, "IPv6 not added yet")
+	assert.NotNil(t, iceBind.udpMux, "mux should be created after first connection")
+	iceBind.muUDPMux.Unlock()
+
+	// Simulate wireguard-go calling CreateReceiverFn for IPv6
+	ipv6RecvFn := rc.CreateReceiverFn(ipv6.NewPacketConn(ipv6Conn), ipv6Conn, false, pool)
+	require.NotNil(t, ipv6RecvFn)
+
+	iceBind.muUDPMux.Lock()
+	assert.NotNil(t, iceBind.ipv4Conn, "should still have IPv4 connection")
+	assert.NotNil(t, iceBind.ipv6Conn, "should now have IPv6 connection")
+	assert.NotNil(t, iceBind.udpMux, "mux should still exist")
+	iceBind.muUDPMux.Unlock()
+
+	mux, err := iceBind.GetICEMux()
+	require.NoError(t, err)
+	require.NotNil(t, mux)
+}
+
+func TestICEBind_WorksWithIPv4Only(t *testing.T) {
+	iceBind := setupICEBind(t)
+
+	ipv4Conn, err := net.ListenUDP("udp4", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	require.NoError(t, err)
+	defer ipv4Conn.Close()
+
+	rc := receiverCreator{iceBind}
+	recvFn := rc.CreateReceiverFn(ipv4.NewPacketConn(ipv4Conn), ipv4Conn, false, createMsgPool())
+	require.NotNil(t, recvFn)
+
+	iceBind.muUDPMux.Lock()
+	assert.NotNil(t, iceBind.ipv4Conn)
+	assert.Nil(t, iceBind.ipv6Conn)
+	assert.NotNil(t, iceBind.udpMux)
+	iceBind.muUDPMux.Unlock()
+
+	mux, err := iceBind.GetICEMux()
+	require.NoError(t, err)
+	require.NotNil(t, mux)
+}
+
+func TestICEBind_WorksWithIPv6Only(t *testing.T) {
+	iceBind := setupICEBind(t)
+
+	ipv6Conn, err := net.ListenUDP("udp6", &net.UDPAddr{IP: net.IPv6zero, Port: 0})
+	if err != nil {
+		t.Skipf("IPv6 not available: %v", err)
+	}
+	defer ipv6Conn.Close()
+
+	rc := receiverCreator{iceBind}
+	recvFn := rc.CreateReceiverFn(ipv6.NewPacketConn(ipv6Conn), ipv6Conn, false, createMsgPool())
+	require.NotNil(t, recvFn)
+
+	iceBind.muUDPMux.Lock()
+	assert.Nil(t, iceBind.ipv4Conn)
+	assert.NotNil(t, iceBind.ipv6Conn)
+	assert.NotNil(t, iceBind.udpMux)
+	iceBind.muUDPMux.Unlock()
+
+	mux, err := iceBind.GetICEMux()
+	require.NoError(t, err)
+	require.NotNil(t, mux)
+}
+
+// TestICEBind_SendsToIPv4AndIPv6PeersSimultaneously verifies that we can communicate
+// with peers on different address families through the same DualStackPacketConn.
+func TestICEBind_SendsToIPv4AndIPv6PeersSimultaneously(t *testing.T) {
+	// two "remote peers" listening on different address families
+	ipv4Peer := listenUDP(t, "udp4", "127.0.0.1:0")
+	defer ipv4Peer.Close()
+
+	ipv6Peer, err := net.ListenUDP("udp6", &net.UDPAddr{IP: net.IPv6loopback, Port: 0})
+	if err != nil {
+		t.Skipf("IPv6 not available: %v", err)
+	}
+	defer ipv6Peer.Close()
+
+	// our local dual-stack connection
+	ipv4Local := listenUDP(t, "udp4", "127.0.0.1:0")
+	defer ipv4Local.Close()
+
+	ipv6Local := listenUDP(t, "udp6", "[::1]:0")
+	defer ipv6Local.Close()
+
+	dualStack := NewDualStackPacketConn(ipv4Local, ipv6Local)
+
+	// send to both peers
+	_, err = dualStack.WriteTo([]byte("to-ipv4"), ipv4Peer.LocalAddr())
+	require.NoError(t, err)
+
+	_, err = dualStack.WriteTo([]byte("to-ipv6"), ipv6Peer.LocalAddr())
+	require.NoError(t, err)
+
+	// verify IPv4 peer got its packet from the IPv4 socket
+	buf := make([]byte, 100)
+	_ = ipv4Peer.SetReadDeadline(time.Now().Add(time.Second))
+	n, addr, err := ipv4Peer.ReadFrom(buf)
+	require.NoError(t, err)
+	assert.Equal(t, "to-ipv4", string(buf[:n]))
+	assert.Equal(t, ipv4Local.LocalAddr().(*net.UDPAddr).Port, addr.(*net.UDPAddr).Port)
+
+	// verify IPv6 peer got its packet from the IPv6 socket
+	_ = ipv6Peer.SetReadDeadline(time.Now().Add(time.Second))
+	n, addr, err = ipv6Peer.ReadFrom(buf)
+	require.NoError(t, err)
+	assert.Equal(t, "to-ipv6", string(buf[:n]))
+	assert.Equal(t, ipv6Local.LocalAddr().(*net.UDPAddr).Port, addr.(*net.UDPAddr).Port)
+}
+
+// TestICEBind_HandlesConcurrentMixedTraffic sends packets concurrently to both IPv4
+// and IPv6 peers. Verifies no packets get misrouted (IPv4 peer only gets v4- packets,
+// IPv6 peer only gets v6- packets). Some packet loss is acceptable for UDP.
+func TestICEBind_HandlesConcurrentMixedTraffic(t *testing.T) {
+	ipv4Peer := listenUDP(t, "udp4", "127.0.0.1:0")
+	defer ipv4Peer.Close()
+
+	ipv6Peer, err := net.ListenUDP("udp6", &net.UDPAddr{IP: net.IPv6loopback, Port: 0})
+	if err != nil {
+		t.Skipf("IPv6 not available: %v", err)
+	}
+	defer ipv6Peer.Close()
+
+	ipv4Local := listenUDP(t, "udp4", "127.0.0.1:0")
+	defer ipv4Local.Close()
+
+	ipv6Local := listenUDP(t, "udp6", "[::1]:0")
+	defer ipv6Local.Close()
+
+	dualStack := NewDualStackPacketConn(ipv4Local, ipv6Local)
+
+	const packetsPerFamily = 500
+
+	ipv4Received := make(chan string, packetsPerFamily)
+	ipv6Received := make(chan string, packetsPerFamily)
+
+	startGate := make(chan struct{})
+	var wg sync.WaitGroup
+
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		buf := make([]byte, 100)
+		for i := 0; i < packetsPerFamily; i++ {
+			n, _, err := ipv4Peer.ReadFrom(buf)
+			if err != nil {
+				return
+			}
+			ipv4Received <- string(buf[:n])
+		}
+	}()
+
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		buf := make([]byte, 100)
+		for i := 0; i < packetsPerFamily; i++ {
+			n, _, err := ipv6Peer.ReadFrom(buf)
+			if err != nil {
+				return
+			}
+			ipv6Received <- string(buf[:n])
+		}
+	}()
+
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		<-startGate
+		for i := 0; i < packetsPerFamily; i++ {
+			_, _ = dualStack.WriteTo([]byte(fmt.Sprintf("v4-%04d", i)), ipv4Peer.LocalAddr())
+		}
+	}()
+
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		<-startGate
+		for i := 0; i < packetsPerFamily; i++ {
+			_, _ = dualStack.WriteTo([]byte(fmt.Sprintf("v6-%04d", i)), ipv6Peer.LocalAddr())
+		}
+	}()
+
+	close(startGate)
+
+	time.AfterFunc(5*time.Second, func() {
+		_ = ipv4Peer.SetReadDeadline(time.Now())
+		_ = ipv6Peer.SetReadDeadline(time.Now())
+	})
+
+	wg.Wait()
+	close(ipv4Received)
+	close(ipv6Received)
+
+	ipv4Count := 0
+	for pkt := range ipv4Received {
+		require.True(t, len(pkt) >= 3 && pkt[:3] == "v4-", "IPv4 peer got misrouted packet: %s", pkt)
+		ipv4Count++
+	}
+
+	ipv6Count := 0
+	for pkt := range ipv6Received {
+		require.True(t, len(pkt) >= 3 && pkt[:3] == "v6-", "IPv6 peer got misrouted packet: %s", pkt)
+		ipv6Count++
+	}
+
+	assert.Equal(t, packetsPerFamily, ipv4Count)
+	assert.Equal(t, packetsPerFamily, ipv6Count)
+}
+
+func TestICEBind_DetectsAddressFamilyFromConnection(t *testing.T) {
+	tests := []struct {
+		name     string
+		network  string
+		addr     string
+		wantIPv4 bool
+	}{
+		{"IPv4 any", "udp4", "0.0.0.0:0", true},
+		{"IPv4 loopback", "udp4", "127.0.0.1:0", true},
+		{"IPv6 any", "udp6", "[::]:0", false},
+		{"IPv6 loopback", "udp6", "[::1]:0", false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			addr, err := net.ResolveUDPAddr(tt.network, tt.addr)
+			require.NoError(t, err)
+
+			conn, err := net.ListenUDP(tt.network, addr)
+			if err != nil {
+				t.Skipf("%s not available: %v", tt.network, err)
+			}
+			defer conn.Close()
+
+			localAddr := conn.LocalAddr().(*net.UDPAddr)
+			isIPv4 := localAddr.IP.To4() != nil
+			assert.Equal(t, tt.wantIPv4, isIPv4)
+		})
+	}
+}
+
+// helpers
+
+func setupICEBind(t *testing.T) *ICEBind {
+	t.Helper()
+	transportNet, err := stdnet.NewNet()
+	require.NoError(t, err)
+
+	address := wgaddr.Address{
+		IP:      netip.MustParseAddr("100.64.0.1"),
+		Network: netip.MustParsePrefix("100.64.0.0/10"),
+	}
+	return NewICEBind(transportNet, nil, address, 1280)
+}
+
+func createDualStackConns(t *testing.T) (*net.UDPConn, *net.UDPConn) {
+	t.Helper()
+	ipv4Conn, err := net.ListenUDP("udp4", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	require.NoError(t, err)
+
+	ipv6Conn, err := net.ListenUDP("udp6", &net.UDPAddr{IP: net.IPv6zero, Port: 0})
+	if err != nil {
+		ipv4Conn.Close()
+		t.Skipf("IPv6 not available: %v", err)
+	}
+	return ipv4Conn, ipv6Conn
+}
+
+func createMsgPool() *sync.Pool {
+	return &sync.Pool{
+		New: func() any {
+			msgs := make([]ipv6.Message, 1)
+			for i := range msgs {
+				msgs[i].Buffers = make(net.Buffers, 1)
+				msgs[i].OOB = make([]byte, 0, 40)
+			}
+			return &msgs
+		},
+	}
+}
+
+func listenUDP(t *testing.T, network, addr string) *net.UDPConn {
+	t.Helper()
+	udpAddr, err := net.ResolveUDPAddr(network, addr)
+	require.NoError(t, err)
+	conn, err := net.ListenUDP(network, udpAddr)
+	require.NoError(t, err)
+	return conn
+}

client/iface/configurer/common.go

@@ -3,8 +3,22 @@ package configurer
 import (
 	"net"
 	"net/netip"
+
+	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
 )
 
+// buildPresharedKeyConfig creates a wgtypes.Config for setting a preshared key on a peer.
+// This is a shared helper used by both kernel and userspace configurers.
+func buildPresharedKeyConfig(peerKey wgtypes.Key, psk wgtypes.Key, updateOnly bool) wgtypes.Config {
+	return wgtypes.Config{
+		Peers: []wgtypes.PeerConfig{{
+			PublicKey:    peerKey,
+			PresharedKey: &psk,
+			UpdateOnly:   updateOnly,
+		}},
+	}
+}
+
 func prefixesToIPNets(prefixes []netip.Prefix) []net.IPNet {
 	ipNets := make([]net.IPNet, len(prefixes))
 	for i, prefix := range prefixes {

client/iface/configurer/kernel_unix.go

@@ -15,8 +15,6 @@ import (
 	"github.com/netbirdio/netbird/monotime"
 )
 
-var zeroKey wgtypes.Key
-
 type KernelConfigurer struct {
 	deviceName string
 }
@@ -48,6 +46,18 @@ func (c *KernelConfigurer) ConfigureInterface(privateKey string, port int) error
 	return nil
 }
 
+// SetPresharedKey sets the preshared key for a peer.
+// If updateOnly is true, only updates the existing peer; if false, creates or updates.
+func (c *KernelConfigurer) SetPresharedKey(peerKey string, psk wgtypes.Key, updateOnly bool) error {
+	parsedPeerKey, err := wgtypes.ParseKey(peerKey)
+	if err != nil {
+		return err
+	}
+
+	cfg := buildPresharedKeyConfig(parsedPeerKey, psk, updateOnly)
+	return c.configure(cfg)
+}
+
 func (c *KernelConfigurer) UpdatePeer(peerKey string, allowedIps []netip.Prefix, keepAlive time.Duration, endpoint *net.UDPAddr, preSharedKey *wgtypes.Key) error {
 	peerKeyParsed, err := wgtypes.ParseKey(peerKey)
 	if err != nil {
@@ -279,7 +289,7 @@ func (c *KernelConfigurer) FullStats() (*Stats, error) {
 			TxBytes:       p.TransmitBytes,
 			RxBytes:       p.ReceiveBytes,
 			LastHandshake: p.LastHandshakeTime,
-			PresharedKey:  p.PresharedKey != zeroKey,
+			PresharedKey:  [32]byte(p.PresharedKey),
 		}
 		if p.Endpoint != nil {
 			peer.Endpoint = *p.Endpoint

client/iface/configurer/uapi.go

@@ -5,20 +5,18 @@ package configurer
 import (
 	"net"
 
-	log "github.com/sirupsen/logrus"
 	"golang.zx2c4.com/wireguard/ipc"
 )
 
 func openUAPI(deviceName string) (net.Listener, error) {
 	uapiSock, err := ipc.UAPIOpen(deviceName)
 	if err != nil {
-		log.Errorf("failed to open uapi socket: %v", err)
 		return nil, err
 	}
 
 	listener, err := ipc.UAPIListen(deviceName, uapiSock)
 	if err != nil {
-		log.Errorf("failed to listen on uapi socket: %v", err)
+		_ = uapiSock.Close()
 		return nil, err
 	}
 

client/iface/configurer/usp.go

@@ -22,17 +22,16 @@ import (
 )
 
 const (
-	privateKey                  = "private_key"
-	ipcKeyLastHandshakeTimeSec  = "last_handshake_time_sec"
-	ipcKeyLastHandshakeTimeNsec = "last_handshake_time_nsec"
-	ipcKeyTxBytes               = "tx_bytes"
-	ipcKeyRxBytes               = "rx_bytes"
-	allowedIP                   = "allowed_ip"
-	endpoint                    = "endpoint"
-	fwmark                      = "fwmark"
-	listenPort                  = "listen_port"
-	publicKey                   = "public_key"
-	presharedKey                = "preshared_key"
+	privateKey                 = "private_key"
+	ipcKeyLastHandshakeTimeSec = "last_handshake_time_sec"
+	ipcKeyTxBytes              = "tx_bytes"
+	ipcKeyRxBytes              = "rx_bytes"
+	allowedIP                  = "allowed_ip"
+	endpoint                   = "endpoint"
+	fwmark                     = "fwmark"
+	listenPort                 = "listen_port"
+	publicKey                  = "public_key"
+	presharedKey               = "preshared_key"
 )
 
 var ErrAllowedIPNotFound = fmt.Errorf("allowed IP not found")
@@ -55,6 +54,14 @@ func NewUSPConfigurer(device *device.Device, deviceName string, activityRecorder
 	return wgCfg
 }
 
+func NewUSPConfigurerNoUAPI(device *device.Device, deviceName string, activityRecorder *bind.ActivityRecorder) *WGUSPConfigurer {
+	return &WGUSPConfigurer{
+		device:           device,
+		deviceName:       deviceName,
+		activityRecorder: activityRecorder,
+	}
+}
+
 func (c *WGUSPConfigurer) ConfigureInterface(privateKey string, port int) error {
 	log.Debugf("adding Wireguard private key")
 	key, err := wgtypes.ParseKey(privateKey)
@@ -72,6 +79,18 @@ func (c *WGUSPConfigurer) ConfigureInterface(privateKey string, port int) error
 	return c.device.IpcSet(toWgUserspaceString(config))
 }
 
+// SetPresharedKey sets the preshared key for a peer.
+// If updateOnly is true, only updates the existing peer; if false, creates or updates.
+func (c *WGUSPConfigurer) SetPresharedKey(peerKey string, psk wgtypes.Key, updateOnly bool) error {
+	parsedPeerKey, err := wgtypes.ParseKey(peerKey)
+	if err != nil {
+		return err
+	}
+
+	cfg := buildPresharedKeyConfig(parsedPeerKey, psk, updateOnly)
+	return c.device.IpcSet(toWgUserspaceString(cfg))
+}
+
 func (c *WGUSPConfigurer) UpdatePeer(peerKey string, allowedIps []netip.Prefix, keepAlive time.Duration, endpoint *net.UDPAddr, preSharedKey *wgtypes.Key) error {
 	peerKeyParsed, err := wgtypes.ParseKey(peerKey)
 	if err != nil {
@@ -100,7 +119,7 @@ func (c *WGUSPConfigurer) UpdatePeer(peerKey string, allowedIps []netip.Prefix,
 		if err != nil {
 			return fmt.Errorf("failed to parse endpoint address: %w", err)
 		}
-		addrPort := netip.AddrPortFrom(addr, uint16(endpoint.Port))
+		addrPort := netip.AddrPortFrom(addr.Unmap(), uint16(endpoint.Port))
 		c.activityRecorder.UpsertAddress(peerKey, addrPort)
 	}
 	return nil
@@ -422,23 +441,19 @@ func toWgUserspaceString(wgCfg wgtypes.Config) string {
 		hexKey := hex.EncodeToString(p.PublicKey[:])
 		sb.WriteString(fmt.Sprintf("public_key=%s\n", hexKey))
 
+		if p.Remove {
+			sb.WriteString("remove=true\n")
+		}
+
+		if p.UpdateOnly {
+			sb.WriteString("update_only=true\n")
+		}
+
 		if p.PresharedKey != nil {
 			preSharedHexKey := hex.EncodeToString(p.PresharedKey[:])
 			sb.WriteString(fmt.Sprintf("preshared_key=%s\n", preSharedHexKey))
 		}
 
-		if p.Remove {
-			sb.WriteString("remove=true")
-		}
-
-		if p.ReplaceAllowedIPs {
-			sb.WriteString("replace_allowed_ips=true\n")
-		}
-
-		for _, aip := range p.AllowedIPs {
-			sb.WriteString(fmt.Sprintf("allowed_ip=%s\n", aip.String()))
-		}
-
 		if p.Endpoint != nil {
 			sb.WriteString(fmt.Sprintf("endpoint=%s\n", p.Endpoint.String()))
 		}
@@ -446,6 +461,14 @@ func toWgUserspaceString(wgCfg wgtypes.Config) string {
 		if p.PersistentKeepaliveInterval != nil {
 			sb.WriteString(fmt.Sprintf("persistent_keepalive_interval=%d\n", int(p.PersistentKeepaliveInterval.Seconds())))
 		}
+
+		if p.ReplaceAllowedIPs {
+			sb.WriteString("replace_allowed_ips=true\n")
+		}
+
+		for _, aip := range p.AllowedIPs {
+			sb.WriteString(fmt.Sprintf("allowed_ip=%s\n", aip.String()))
+		}
 	}
 	return sb.String()
 }
@@ -543,7 +566,7 @@ func parseStatus(deviceName, ipcStr string) (*Stats, error) {
 				continue
 			}
 
-			host, portStr, err := net.SplitHostPort(strings.Trim(val, "[]"))
+			host, portStr, err := net.SplitHostPort(val)
 			if err != nil {
 				log.Errorf("failed to parse endpoint: %v", err)
 				continue
@@ -599,7 +622,9 @@ func parseStatus(deviceName, ipcStr string) (*Stats, error) {
 				continue
 			}
 			if val != "" && val != "0000000000000000000000000000000000000000000000000000000000000000" {
-				currentPeer.PresharedKey = true
+				if pskKey, err := hexToWireguardKey(val); err == nil {
+					currentPeer.PresharedKey = [32]byte(pskKey)
+				}
 			}
 		}
 	}

client/iface/configurer/wgshow.go

@@ -12,7 +12,7 @@ type Peer struct {
 	TxBytes       int64
 	RxBytes       int64
 	LastHandshake time.Time
-	PresharedKey  bool
+	PresharedKey  [32]byte
 }
 
 type Stats struct {

client/iface/device/adapter.go

@@ -2,7 +2,7 @@ package device
 
 // TunAdapter is an interface for create tun device from external service
 type TunAdapter interface {
-	ConfigureInterface(address string, mtu int, dns string, searchDomains string, routes string) (int, error)
+	ConfigureInterface(address string, addressV6 string, mtu int, dns string, searchDomains string, routes string) (int, error)
 	UpdateAddr(address string) error
 	ProtectSocket(fd int32) bool
 }

client/iface/device/device_android.go

@@ -63,7 +63,7 @@ func (t *WGTunDevice) Create(routes []string, dns string, searchDomains []string
 		searchDomainsToString = ""
 	}
 
-	fd, err := t.tunAdapter.ConfigureInterface(t.address.String(), int(t.mtu), dns, searchDomainsToString, routesString)
+	fd, err := t.tunAdapter.ConfigureInterface(t.address.String(), t.address.IPv6String(), int(t.mtu), dns, searchDomainsToString, routesString)
 	if err != nil {
 		log.Errorf("failed to create Android interface: %s", err)
 		return nil, err

client/iface/device/device_darwin.go

@@ -131,23 +131,32 @@ func (t *TunDevice) Device() *device.Device {
 
 // assignAddr Adds IP address to the tunnel interface and network route based on the range provided
 func (t *TunDevice) assignAddr() error {
-	cmd := exec.Command("ifconfig", t.name, "inet", t.address.IP.String(), t.address.IP.String())
-	if out, err := cmd.CombinedOutput(); err != nil {
-		log.Errorf("adding address command '%v' failed with output: %s", cmd.String(), out)
-		return err
+	if out, err := exec.Command("ifconfig", t.name, "inet", t.address.IP.String(), t.address.IP.String()).CombinedOutput(); err != nil {
+		return fmt.Errorf("add v4 address: %s: %w", string(out), err)
 	}
 
-	// dummy ipv6 so routing works
-	cmd = exec.Command("ifconfig", t.name, "inet6", "fe80::/64")
-	if out, err := cmd.CombinedOutput(); err != nil {
-		log.Debugf("adding address command '%v' failed with output: %s", cmd.String(), out)
+	// Assign a dummy link-local so macOS enables IPv6 on the tun device.
+	// When a real overlay v6 is present, use that instead.
+	v6Addr := "fe80::/64"
+	if t.address.HasIPv6() {
+		v6Addr = t.address.IPv6String()
+	}
+	if out, err := exec.Command("ifconfig", t.name, "inet6", v6Addr).CombinedOutput(); err != nil {
+		log.Warnf("failed to assign IPv6 address %s, continuing v4-only: %s: %v", v6Addr, string(out), err)
+		t.address.ClearIPv6()
 	}
 
-	routeCmd := exec.Command("route", "add", "-net", t.address.Network.String(), "-interface", t.name)
-	if out, err := routeCmd.CombinedOutput(); err != nil {
-		log.Errorf("adding route command '%v' failed with output: %s", routeCmd.String(), out)
-		return err
+	if out, err := exec.Command("route", "add", "-net", t.address.Network.String(), "-interface", t.name).CombinedOutput(); err != nil {
+		return fmt.Errorf("add route %s via %s: %s: %w", t.address.Network, t.name, string(out), err)
 	}
+
+	if t.address.HasIPv6() {
+		if out, err := exec.Command("route", "add", "-inet6", "-net", t.address.IPv6Net.String(), "-interface", t.name).CombinedOutput(); err != nil {
+			log.Warnf("failed to add route %s via %s, continuing v4-only: %s: %v", t.address.IPv6Net, t.name, string(out), err)
+			t.address.ClearIPv6()
+		}
+	}
+
 	return nil
 }
 

client/iface/device/device_filter.go

@@ -29,8 +29,9 @@ type PacketFilter interface {
 type FilteredDevice struct {
 	tun.Device
 
-	filter PacketFilter
-	mutex  sync.RWMutex
+	filter    PacketFilter
+	mutex     sync.RWMutex
+	closeOnce sync.Once
 }
 
 // newDeviceFilter constructor function
@@ -40,6 +41,20 @@ func newDeviceFilter(device tun.Device) *FilteredDevice {
 	}
 }
 
+// Close closes the underlying tun device exactly once.
+// wireguard-go's netTun.Close() panics on double-close due to a bare close(channel),
+// and multiple code paths can trigger Close on the same device.
+func (d *FilteredDevice) Close() error {
+	var err error
+	d.closeOnce.Do(func() {
+		err = d.Device.Close()
+	})
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
 // Read wraps read method with filtering feature
 func (d *FilteredDevice) Read(bufs [][]byte, sizes []int, offset int) (n int, err error) {
 	if n, err = d.Device.Read(bufs, sizes, offset); err != nil {

client/iface/device/device_ios.go

@@ -1,6 +1,3 @@
-//go:build ios
-// +build ios
-
 package device
 
 import (
@@ -154,8 +151,11 @@ func (t *TunDevice) MTU() uint16 {
 	return t.mtu
 }
 
-func (t *TunDevice) UpdateAddr(_ wgaddr.Address) error {
-	// todo implement
+// UpdateAddr updates the device address. On iOS the tunnel is managed by the
+// NetworkExtension, so we only store the new value. The extension picks up the
+// change on the next tunnel reconfiguration.
+func (t *TunDevice) UpdateAddr(addr wgaddr.Address) error {
+	t.address = addr
 	return nil
 }