-
Bug
-
Resolution: Unresolved
-
Critical
-
None
-
4.13.z
-
None
-
Quality / Stability / Reliability
-
False
-
-
1
-
Critical
-
No
-
None
-
None
-
Rejected
-
None
-
None
-
Release Note Not Required
-
N/A
-
None
-
None
-
None
-
None
Description of problem:
https://access.redhat.com/solutions/7014337 was reported for RHEL8, but it can happen even in RHEL9.
Bug 2208905[1] - Regression of 3b8cc6298724 ("blk-cgroup: Optimize blkcg_rstat_flush()")
This issue can happen also on OCP 4.13.z.
Once OCPBUGS-17958[3] was filed to track bug 2208905[1](blk-cgorup), but the tracking bug was changed from 2208905[2](blk-cgroup) to 2233214[3](percpu).
Therefore, to track bug 2208905[1](blk-cgorup), I file this OCPBUGS.
Note: When a new bug for RHEL9.2 on this issue, the bug number which we have to track may be changed.
[1] https://bugzilla.redhat.com/show_bug.cgi?id=2208905
[2] https://bugzilla.redhat.com/show_bug.cgi?id=2233214
[3] https://issues.redhat.com/browse/OCPBUGS-17958
Version-Release number of selected component (if applicable):
4.13.6
How reproducible:
Just create and delete pods repeateadly for a while, then check /proc/cgroups. You will see that blkio's num_cgroups has increased.
Steps to Reproduce:
1. 2. 3.
Actual results:
You will see that blkio's num_cgroups has increased as follows: ``` [core@worker2 ~]$ cat /proc/cgroups #subsys_name hierarchy num_cgroups enabled cpuset 13 70 1 cpu 12 119 1 cpuacct 12 119 1 blkio 2 4689 1 memory 3 265 1 devices 4 101 1 freezer 9 70 1 net_cls 5 70 1 perf_event 8 70 1 net_prio 5 70 1 hugetlb 11 70 1 pids 7 120 1 rdma 6 70 1 misc 10 70 1 ```
Expected results:
It's not increased.
Additional info: