Files
netbird/client/internal/routemanager/dynamic/route.go
Riccardo Manfrin 2f268c8141 [client] fix stale routing peer on overlapping-prefix network removal (#6799)
## Describe your changes

The AllowedIPs reference counter
([refcounter/types.go#L9](https://github.com/netbirdio/netbird/blob/e1a24376a/client/internal/routemanager/refcounter/types.go#L9))
was keyed only by prefix and stored a single active peer set by
the first registrar, never swapped. When two networks advertised the
same prefix via different routing peers, removing the one whose peer was
installed in WireGuard left the prefix pointing at the removed peer
instead of the
surviving one — traffic kept flowing to the old peer until a manual
`netbird down/up`.

Made the AllowedIPs counter peer-aware: it tracks a per-peer reference
count per prefix plus the installed peer, and swaps WireGuard to a
surviving peer when the active one releases its last reference (removes
the prefix when
none remain). `Decrement` now takes the peer key so the exact
incremented peer is released; the static handler records its selected
routing peer like the dynamic and DNS handlers already did. The generic
`Counter` (routes,
  exclusion, ipset) is unchanged.

  ## Issue ticket number and link

No public issue — reported internally (routes not updating without
`netbird down/up` when two networks share a subnet). Root cause is the
prefix-only key at

[refcounter/types.go#L9](https://github.com/netbirdio/netbird/blob/e1a24376a/client/internal/routemanager/refcounter/types.go#L9).

  ## Stack

  <!-- branch-stack -->

  ### Checklist
  - [x] Is it a bug fix
  - [ ] Is a typo/documentation fix
  - [ ] Is a feature enhancement
  - [ ] It is a refactor
  - [x] Created tests that fail without the change (if possible)

> By submitting this pull request, you confirm that you have read and
agree to the terms of the [Contributor License
Agreement](https://github.com/netbirdio/netbird/blob/main/CONTRIBUTOR_LICENSE_AGREEMENT.md).

  ## Documentation
  Select exactly one:

  - [ ] I added/updated documentation for this change
  - [x] Documentation is **not needed** for this change (explain why)

Internal client-side routing fix. No public API, CLI, or config change —
only the WireGuard AllowedIPs hand-off when overlapping-prefix networks
are removed.

  ### Docs PR URL (required if "docs added" is checked)
  Paste the PR link from https://github.com/netbirdio/docs here:

  N/A

<!-- codesmith:footer -->
---
<a
href="https://app.blacksmith.sh/netbirdio/codesmith/netbird/pr/6799"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://pr-comments-assets.blacksmith.sh/codesmith/view-with-codesmith-dark-v2.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://pr-comments-assets.blacksmith.sh/codesmith/view-with-codesmith-light-v2.svg"><img
alt="View with Codesmith"
src="https://pr-comments-assets.blacksmith.sh/codesmith/view-with-codesmith-dark-v2.svg"></picture></a>
<a
href="https://backend.blacksmith.sh/track/enable-autofix?expires=1786796229&installation_id=146802194&pr_number=6799&repository=netbirdio%2Fnetbird&return_to=https%3A%2F%2Fgithub.com%2Fnetbirdio%2Fnetbird%2Fpull%2F6799&signature=322d6950b4f664b1cb3421f2efa039fbbb7946de9b79f2a1b1b131dc182ada2c"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://pr-comments-assets.blacksmith.sh/codesmith/autofix-with-codesmith-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://pr-comments-assets.blacksmith.sh/codesmith/autofix-with-codesmith-light.svg"><img
alt="Autofix with Codesmith"
src="https://pr-comments-assets.blacksmith.sh/codesmith/autofix-with-codesmith-dark.svg"></picture></a>
<sup>Need help on this PR? Tag <code>/codesmith</code> with what you
need. Autofix is disabled.</sup>

<!-- codesmith:autofix:disabled -->
<!-- /codesmith:footer -->

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **Bug Fixes**
* Improved routing behavior when multiple peers share the same Allowed
IP by making Allowed IP reference tracking peer-aware.
* Allowed IPs now correctly decrement using the active peer key and
transfer to another surviving active peer when the current peer is
removed.
* Prevented stale routing and incorrect reference cleanup during route
and DNS-driven teardown.

* **Tests**
* Added/extended coverage for peer handoffs, repeated references,
non-active peer removal, flushing behavior, and self-healing after swap
add/remove failures.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-07-28 14:10:58 +09:00

400 lines
11 KiB
Go

package dynamic
import (
"context"
"fmt"
"net"
"net/netip"
"strings"
"sync"
"time"
"github.com/hashicorp/go-multierror"
log "github.com/sirupsen/logrus"
nberrors "github.com/netbirdio/netbird/client/errors"
"github.com/netbirdio/netbird/client/internal/peer"
"github.com/netbirdio/netbird/client/internal/routemanager/common"
"github.com/netbirdio/netbird/client/internal/routemanager/iface"
"github.com/netbirdio/netbird/client/internal/routemanager/refcounter"
"github.com/netbirdio/netbird/client/internal/routemanager/util"
"github.com/netbirdio/netbird/route"
"github.com/netbirdio/netbird/shared/management/domain"
)
const (
DefaultInterval = time.Minute
minInterval = 2 * time.Second
failureInterval = 5 * time.Second
addAllowedIP = "add allowed IP %s: %w"
)
type domainMap map[domain.Domain][]netip.Prefix
type resolveResult struct {
domain domain.Domain
prefix netip.Prefix
err error
}
type Route struct {
route *route.Route
routeRefCounter *refcounter.RouteRefCounter
allowedIPsRefcounter *refcounter.AllowedIPsRefCounter
interval time.Duration
dynamicDomains domainMap
mu sync.Mutex
currentPeerKey string
cancel context.CancelFunc
statusRecorder *peer.Status
wgInterface iface.WGIface
resolverAddr netip.AddrPort
}
func NewRoute(params common.HandlerParams, resolverAddr netip.AddrPort) *Route {
return &Route{
route: params.Route,
routeRefCounter: params.RouteRefCounter,
allowedIPsRefcounter: params.AllowedIPsRefCounter,
interval: params.DnsRouterInterval,
statusRecorder: params.StatusRecorder,
wgInterface: params.WgInterface,
resolverAddr: resolverAddr,
dynamicDomains: domainMap{},
}
}
func (r *Route) String() string {
return r.route.Domains.SafeString()
}
func (r *Route) AddRoute(ctx context.Context) error {
r.mu.Lock()
defer r.mu.Unlock()
if r.cancel != nil {
r.cancel()
}
ctx, r.cancel = context.WithCancel(ctx)
go r.startResolver(ctx)
return nil
}
// RemoveRoute will stop the dynamic resolver and remove all dynamic routes.
// It doesn't touch allowed IPs, these should be removed separately and before calling this method.
func (r *Route) RemoveRoute() error {
r.mu.Lock()
defer r.mu.Unlock()
if r.cancel != nil {
r.cancel()
}
var merr *multierror.Error
for domain, prefixes := range r.dynamicDomains {
for _, prefix := range prefixes {
if _, err := r.routeRefCounter.Decrement(prefix); err != nil {
merr = multierror.Append(merr, fmt.Errorf("remove dynamic route for IP %s: %w", prefix, err))
}
}
log.Debugf("Removed dynamic route(s) for [%s]: %s", domain.SafeString(), strings.ReplaceAll(fmt.Sprintf("%s", prefixes), " ", ", "))
r.statusRecorder.DeleteResolvedDomainsStates(domain)
}
r.dynamicDomains = domainMap{}
return nberrors.FormatErrorOrNil(merr)
}
func (r *Route) AddAllowedIPs(peerKey string) error {
r.mu.Lock()
defer r.mu.Unlock()
var merr *multierror.Error
for domain, domainPrefixes := range r.dynamicDomains {
for _, prefix := range domainPrefixes {
if err := r.incrementAllowedIP(domain, prefix, peerKey); err != nil {
merr = multierror.Append(merr, fmt.Errorf(addAllowedIP, prefix, err))
}
}
}
r.currentPeerKey = peerKey
return nberrors.FormatErrorOrNil(merr)
}
func (r *Route) RemoveAllowedIPs() error {
r.mu.Lock()
defer r.mu.Unlock()
var merr *multierror.Error
for _, domainPrefixes := range r.dynamicDomains {
for _, prefix := range domainPrefixes {
if _, err := r.allowedIPsRefcounter.Decrement(prefix, r.currentPeerKey); err != nil {
merr = multierror.Append(merr, fmt.Errorf("remove allowed IP %s: %w", prefix, err))
}
}
}
r.currentPeerKey = ""
return nberrors.FormatErrorOrNil(merr)
}
func (r *Route) startResolver(ctx context.Context) {
log.Debugf("Starting dynamic route resolver for domains [%v]", r)
interval := r.interval
if interval < minInterval {
interval = minInterval
log.Warnf("Dynamic route resolver interval %s is too low, setting to minimum value %s", r.interval, minInterval)
}
ticker := time.NewTicker(interval)
defer ticker.Stop()
if err := r.update(ctx); err != nil {
log.Errorf("Failed to resolve domains for route [%v]: %v", r, err)
if interval > failureInterval {
ticker.Reset(failureInterval)
}
}
for {
select {
case <-ctx.Done():
log.Debugf("Stopping dynamic route resolver for domains [%v]", r)
return
case <-ticker.C:
if err := r.update(ctx); err != nil {
log.Errorf("Failed to resolve domains for route [%v]: %v", r, err)
// Use a lower ticker interval if the update fails
if interval > failureInterval {
ticker.Reset(failureInterval)
}
} else if interval > failureInterval {
// Reset to the original interval if the update succeeds
ticker.Reset(interval)
}
}
}
}
func (r *Route) update(ctx context.Context) error {
resolved, err := r.resolveDomains(ctx)
if err != nil {
if len(resolved) == 0 {
return fmt.Errorf("resolve domains: %w", err)
}
log.Warnf("Failed to resolve domains: %v", err)
}
if err := r.updateDynamicRoutes(ctx, resolved); err != nil {
return fmt.Errorf("update dynamic routes: %w", err)
}
return nil
}
func (r *Route) resolveDomains(ctx context.Context) (domainMap, error) {
results := make(chan resolveResult)
go r.resolve(ctx, results)
resolved := domainMap{}
var merr *multierror.Error
for result := range results {
if result.err != nil {
merr = multierror.Append(merr, result.err)
} else {
resolved[result.domain] = append(resolved[result.domain], result.prefix)
}
}
return resolved, nberrors.FormatErrorOrNil(merr)
}
func (r *Route) resolve(ctx context.Context, results chan resolveResult) {
var wg sync.WaitGroup
for _, d := range r.route.Domains {
wg.Add(1)
go func(domain domain.Domain) {
defer wg.Done()
ips, err := r.getIPsFromResolver(ctx, domain)
if err != nil {
log.Tracef("Failed to resolve domain %s with private resolver: %v", domain.SafeString(), err)
ips, err = lookupHostIPs(ctx, domain)
if err != nil {
results <- resolveResult{domain: domain, err: fmt.Errorf("resolve d %s: %w", domain.SafeString(), err)}
return
}
}
for _, ip := range ips {
prefix, err := util.GetPrefixFromIP(ip)
if err != nil {
results <- resolveResult{domain: domain, err: fmt.Errorf("get prefix from IP %s: %w", ip.String(), err)}
return
}
results <- resolveResult{domain: domain, prefix: prefix}
}
}(d)
}
wg.Wait()
close(results)
}
func (r *Route) updateDynamicRoutes(ctx context.Context, newDomains domainMap) error {
r.mu.Lock()
defer r.mu.Unlock()
if ctx.Err() != nil {
return ctx.Err()
}
var merr *multierror.Error
for domain, newPrefixes := range newDomains {
oldPrefixes := r.dynamicDomains[domain]
toAdd, toRemove := determinePrefixChanges(oldPrefixes, newPrefixes)
addedPrefixes, err := r.addRoutes(domain, toAdd)
if err != nil {
merr = multierror.Append(merr, err)
} else if len(addedPrefixes) > 0 {
log.Debugf("Added dynamic route(s) for [%s]: %s", domain.SafeString(), strings.ReplaceAll(fmt.Sprintf("%s", addedPrefixes), " ", ", "))
}
removedPrefixes, err := r.removeRoutes(toRemove)
if err != nil {
merr = multierror.Append(merr, err)
} else if len(removedPrefixes) > 0 {
log.Debugf("Removed dynamic route(s) for [%s]: %s", domain.SafeString(), strings.ReplaceAll(fmt.Sprintf("%s", removedPrefixes), " ", ", "))
}
updatedPrefixes := combinePrefixes(oldPrefixes, removedPrefixes, addedPrefixes)
r.dynamicDomains[domain] = updatedPrefixes
r.statusRecorder.UpdateResolvedDomainsStates(domain, domain, updatedPrefixes, r.route.GetResourceID())
}
return nberrors.FormatErrorOrNil(merr)
}
func (r *Route) addRoutes(domain domain.Domain, prefixes []netip.Prefix) ([]netip.Prefix, error) {
var addedPrefixes []netip.Prefix
var merr *multierror.Error
for _, prefix := range prefixes {
if _, err := r.routeRefCounter.Increment(prefix, struct{}{}); err != nil {
merr = multierror.Append(merr, fmt.Errorf("add dynamic route for IP %s: %w", prefix, err))
continue
}
if r.currentPeerKey != "" {
if err := r.incrementAllowedIP(domain, prefix, r.currentPeerKey); err != nil {
merr = multierror.Append(merr, fmt.Errorf(addAllowedIP, prefix, err))
}
}
addedPrefixes = append(addedPrefixes, prefix)
}
return addedPrefixes, merr.ErrorOrNil()
}
func (r *Route) removeRoutes(prefixes []netip.Prefix) ([]netip.Prefix, error) {
if r.route.KeepRoute {
return nil, nil
}
var removedPrefixes []netip.Prefix
var merr *multierror.Error
for _, prefix := range prefixes {
if _, err := r.routeRefCounter.Decrement(prefix); err != nil {
merr = multierror.Append(merr, fmt.Errorf("remove dynamic route for IP %s: %w", prefix, err))
}
if r.currentPeerKey != "" {
if _, err := r.allowedIPsRefcounter.Decrement(prefix, r.currentPeerKey); err != nil {
merr = multierror.Append(merr, fmt.Errorf("remove allowed IP %s: %w", prefix, err))
}
}
removedPrefixes = append(removedPrefixes, prefix)
}
return removedPrefixes, merr.ErrorOrNil()
}
func (r *Route) incrementAllowedIP(domain domain.Domain, prefix netip.Prefix, peerKey string) error {
if ref, err := r.allowedIPsRefcounter.Increment(prefix, peerKey); err != nil {
return fmt.Errorf(addAllowedIP, prefix, err)
} else if ref.Count > 1 && ref.Out != peerKey {
log.Warnf("IP [%s] for domain [%s] is already routed by peer [%s]. HA routing disabled",
prefix.Addr(),
domain.SafeString(),
ref.Out,
)
}
return nil
}
func determinePrefixChanges(oldPrefixes, newPrefixes []netip.Prefix) (toAdd, toRemove []netip.Prefix) {
prefixSet := make(map[netip.Prefix]bool)
for _, prefix := range oldPrefixes {
prefixSet[prefix] = false
}
for _, prefix := range newPrefixes {
if _, exists := prefixSet[prefix]; exists {
prefixSet[prefix] = true
} else {
toAdd = append(toAdd, prefix)
}
}
for prefix, inUse := range prefixSet {
if !inUse {
toRemove = append(toRemove, prefix)
}
}
return
}
// lookupHostIPs resolves d via the system resolver, honoring ctx cancellation.
func lookupHostIPs(ctx context.Context, d domain.Domain) ([]net.IP, error) {
addrs, err := net.DefaultResolver.LookupIPAddr(ctx, d.PunycodeString())
if err != nil {
return nil, err
}
ips := make([]net.IP, 0, len(addrs))
for _, addr := range addrs {
ips = append(ips, addr.IP)
}
return ips, nil
}
func combinePrefixes(oldPrefixes, removedPrefixes, addedPrefixes []netip.Prefix) []netip.Prefix {
prefixSet := make(map[netip.Prefix]struct{})
for _, prefix := range oldPrefixes {
prefixSet[prefix] = struct{}{}
}
for _, prefix := range removedPrefixes {
delete(prefixSet, prefix)
}
for _, prefix := range addedPrefixes {
prefixSet[prefix] = struct{}{}
}
var combinedPrefixes []netip.Prefix
for prefix := range prefixSet {
combinedPrefixes = append(combinedPrefixes, prefix)
}
return combinedPrefixes
}