-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.14
-
Moderate
-
No
-
False
-
-
Previously, the must-gather tool collected CPU-related performance data sequentially per node, leading to significantly longer collection times in larger clusters. The data is now collected in parallel, resulting in a reduction in collection time.
-
Bug Fix
-
Done
-
-
-
-
This is a clone of issue OCPBUGS-34360. The following is the description of the original issue:
—
Description of problem:
After upgrading to OpenShift 4.14, the must-gather took much longer than before.
Version-Release number of selected component (if applicable):
4.14
How reproducible:
Always
Steps to Reproduce:
1. Run oc adm must-gather 2. Wait for it to complete 3.
Actual results:
For a cluster with around 50 nodes, the must-gather took about 30 minutes.
Expected results:
For a cluster with around 50 nodes, the must-gather can finish in about 10 minutes.
Additional info:
It seems the gather_ppc collection script is related here. https://github.com/openshift/must-gather/blob/release-4.14/collection-scripts/gather_ppc
- clones
-
OCPBUGS-34360 The must-gather took too long after upgrading to OpenShift 4.14
- Closed
- is blocked by
-
OCPBUGS-35357 The must-gather took too long after upgrading to OpenShift 4.14
- Closed
- links to