ooom killed ruby2.3 process on instances where fluentd is running.

[Saggi432]

we have been facing the below errors when we run the fluentd on our cluster.. apparently getting killed with oom

on researching it looks like may have to include jmalloc as part of the ruby base image.. can you please confirm if you have any other proposed solutions instead ?

fluent/fluent-bit-kubernetes-logging#14

391524.805414] Memory cgroup out of memory: Kill process 3685518 (ruby2.3) score 1792 or sacrifice child
[468138.916915] Memory cgroup out of memory: Kill process 859933 (ruby2.3) score 1792 or sacrifice child
[647901.248286] Memory cgroup out of memory: Kill process 2442358 (ruby2.3) score 1792 or sacrifice child
[675732.117926] IPv6: ADDRCONF(NETDEV_UP): vethb0d99b50: link is not ready
[729039.765037] [371792] 0 371792 164959 39926 266 4 0 994 ruby2.3
[729039.774334] Memory cgroup out of memory: Kill process 371792 (ruby2.3) score 1775 or sacrifice child
[729039.783915] Killed process 371792 (ruby2.3) total-vm:659836kB, anon-rss:149212kB, file-rss:10492kB, shmem-rss:0kB
[729039.821620] oom_reaper: reaped process 371792 (ruby2.3), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
[853128.254617] Memory cgroup out of memory: Kill process 2316873 (ruby2.3) score 1792 or sacrifice child
[953790.700466] Memory cgroup out of memory: Kill process 3246659 (ruby2.3) score 1792 or sacrifice child`

[idohalevi]

@Saggi432
Were you able to solve the issue? We never had this issue before

[bdashrad]

Try raising the memory requests and limits in your deployment. The lines here can be changed https://github.com/logzio/logzio-k8s/blob/master/logzio-daemonset.yaml#L29-L34

[idohalevi]

@bdashrad
Great feedback!! thank you for the reply