-
Bug
-
Resolution: Cannot Reproduce
-
Undefined
-
None
-
4.12
-
None
-
OCPNODE Sprint 228 (Blue), OCPNODE Sprint 229 (Blue)
-
2
-
False
-
-
-
Description of problem:
Time to become ready of a single deployment with 60 replicas increases from ~3 minutes on the 1st iteration to ~10 minutes when deleting/re-creating the deployment multiple times
Version-Release number of selected component (if applicable):
4.12.0-rc.0
How reproducible:
100%
Steps to Reproduce:
1. Create the attached deployment 2. Delete and re-create 100 times
Actual results:
We can notice that on the 1st iteration it takes around 3 minutes until all pods become ready. On the 37th create iteration it takes around 10 minutes util all pods become ready.
Expected results:
Time until pods become ready remains constant over the delete/create iterations.
Additional info:
Attaching deployment, must-gather and a log file watching the pods status on every second. Note that the test ran on an SNO cluster with Telco DU profile, with workload partitioning enabled, using 2 cores for the platform components.