Description of problem (please be detailed as possible and provide log snippests): RHEL 9 (cgroups v2) - the pid limits ARE enforced as compared to RHEL8 (cgroup v1) Version of all relevant components (if applicable): RHCS built on RHEL9, RHCS6 and above! For more information please check the RHCS6 bug - https://bugzilla.redhat.com/show_bug.cgi?id=2165644#c12
(In reply to Travis Nielsen from comment #8) > Before we move it over, it would be good to know if there could be a repro > in OCP, or perhaps the limits are already higher in OCP. > Orit, suggestions on an OCP contact? Using --pids-limit -1 is not supported in OCP (from 4.8 if I recall correctly). as it basically removes the ability to limit the PIDs generate by the pod. See: https://bugzilla.redhat.com/show_bug.cgi?id=2039187 I agree and recommend not using it for Ceph containers. ODF uses the default RGW thread pool size so the limit of 1024 should suffice. We will need to document for users increasing that they will need o update the pids limits in OCP accordingly.
See https://access.redhat.com/solutions/5305611 for more info on how to increase the pids limit in OCP
Thanks Orit, moving over to documentation.