Network & Infrastructure / Linux Networking & Host Performance
Bond failover MAC behavior and stale upstream learning
An active-backup bond switches to its standby link, but traffic takes too long to recover. Upstream forwarding or peer neighbor state may still point at the previous attachment.
Environment and inputs: Linux network namespaces, host routing and network drivers; availability and semantics depend on kernel and driver versions. Diagnostic/design example; not tested on production equipment.
Collect the evidence
Record bond MAC policy, active slave changes, peer notifications, switch FDB movement and neighbor cache updates. Compare host failover time with the actual upstream learning delay.
Explain the result
A local active-slave change is only one part of recovery. The surrounding network must learn the correct attachment and accept the bond's MAC behavior according to the selected mode.
Plan the fix
Choose documented bond MAC and notification behavior compatible with the switch and virtualization design. Investigate security features that reject legitimate movement before increasing notification traffic.
Check the result and rollback
Measure end-to-end recovery for each slave loss and restoration. Revert if the new policy creates duplicate MAC ownership or causes port-security events during ordinary failover.