-
Bug
-
Resolution: Unresolved
-
Critical
-
4.19.z, 4.20.0
Description of problem:
We observe cluster-image-registry-operator pods using large number of inotify instances running into /proc/sys/fs/inotify/max_user_instances limit eventually.
Version-Release number of selected component (if applicable):
openshift-v4.19.0 Since the only cluster-image-registry-operator that is not leaking inotify instances is the one from an 4.18.1 cluster, the problem seems to be introduced in 4.19
How reproducible:
100% (in ARO HCP environment)
Steps to Reproduce:
1. Install ARO HCP service infrastructure and provision few clusters 2. Notice issues with image-registry because of too many open files
Actual results:
Expected results:
Additional info:
See also: * https://redhat-external.slack.com/archives/C075PHEFZKQ/p1753170500903469?thread_ts=1752842640.522769&cid=C075PHEFZKQ * https://issues.redhat.com/browse/ARO-19873
- blocks
-
OCPBUGS-60517 Image registry operator is leaking inotify file descriptors
-
- Closed
-
- is cloned by
-
OCPBUGS-60517 Image registry operator is leaking inotify file descriptors
-
- Closed
-
- links to