I couldn't access worker nodes, but on a master node, the ps -elf indicated rmc had 10000 pages. That's ~640 MBs on mostly idle servers - master nodes. The failure above occurred on worker nodes, where ocs is deployed and I am running ocs-ci performance suite which deploys a bunch of fio pods...
The limit should be increased from 1 Gi to 1.5 or 2 Gi.
OOM occurred on worker-1 node. rmcd process targeted.
Events: Type Reason Age From Message
Warning SystemOOM 8h kubelet System OOM encountered, victim process: rmcd, pid: 6150 Warning SystemOOM 8h kubelet System OOM encountered, victim process: rmcd, pid: 1483512 Warning SystemOOM 8h kubelet System OOM encountered, victim process: rmcd, pid: 1486672 Warning SystemOOM 8h kubelet System OOM encountered, victim process: rmcd, pid: 1486690
I couldn't access worker nodes, but on a master node, the ps -elf indicated rmc had 10000 pages. That's ~640 MBs on mostly idle servers - master nodes. The failure above occurred on worker nodes, where ocs is deployed and I am running ocs-ci performance suite which deploys a bunch of fio pods...
The limit should be increased from 1 Gi to 1.5 or 2 Gi.