[BUG] OOM on ip-control-loop

[talcoh2x]

Describe the bug with k8s 1.22 I'm able to see/get more and more issues like Memory cgroup out of memory: Killed process 643039 (ip-control-loop) total-vm:742792kB, anon-rss:43452kB, file-rss:0kB, shmem-rss:0kB, UID:0 pgtables:240kB oom_score_adj:-997

this error does not seem all the time and I'm also not sure about the impact of that on the PODS. there is something I can check to understand from where it comes?

Environment:

Limits:
  cpu:     100m
  memory:  50Mi
Requests:
  cpu:     100m
  memory:  50Mi

• Whereabouts version : Image: ghcr.io/k8snetworkplumbingwg/whereabouts:latest-amd64 can i get the specific version somehow?

[talcoh2x]

[talcoh2x]

from one of the Nodes that I got OOM. what can increase the Memory like that ?

[maiqueb]

from one of the Nodes that I got OOM. what can increase the Memory like that ?

I'm unsure, but, we're still using the requests / limits of when ~~this~~ whereabouts did not have a dedicated controller - since the whereabouts pods now run a controller that reconciles the IP addresses based on a cron expr, and also makes sure the IPs are gone when pods are deleted, it is bound to require more memory.

[dougbtv]

It's possible that the spike is from teh reconciliation being scheduled: https://github.com/k8snetworkplumbingwg/whereabouts/pull/238