Skip to content

Feat: preserve vip on leadership loss

Placeholder Brian Davis requested to merge github/fork/slimm609/preserve_vip_leadership into main

When in arp mode, when preserveVipOnLeadershipLoss is true, when a node loses leadership (or can't keep it), this will stop performing ARPs but will wait until a new leader is elected before deleting the arp. This is beneficial for cases where there may be k8s api or etcd instability.

When the flag is set to false (default) it continue to function like normal.

Current behavior:

  • Leader loses election
  • Deletes VIP immediately <- VIP becomes unavailable
  • New leader elected
  • Configure VIP
  • Starts ARPing <- VIP becomes available again

New flag behavior:

  • Leader loses election
  • Stops ARPing BUT keeps VIP <- vip still on current worker
  • New leader elected
  • Configure VIP (or keeps it if the same node is elected)
  • Starts ARPing <- takes over traffic
  • Cleans up old vip (if on a different kube-vip)

Merge request reports