-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.16.0
-
Important
-
Yes
-
False
-
Description of problem:
Noticed increase in podLatency metrics in 4.16 Rosa HCP clusters compared to 4.15 HCP.
Version-Release number of selected component (if applicable):
4.16.0-0.nightly-2024-06-27-091410
How reproducible:
Rerunning tests to see if we can consistently reproduce.
Steps to Reproduce:
1. Run node-density-heavy tests on 24 nodes
2.
3.
Actual results:
podLatency higher than expected, when comparing with 4.15
Consolidating all the node-density-heavy runs so far:
| 4.16 | worker size | Ready | |||
| 50th | 99th | max | avg | ||
| m5.xlarge | 30000 | 57000 | 150000 | 20422 | |
| m5.xlarge | 15000 | 42000 | 70000 | 19956 | |
| m5.xlarge | 12000 | 42000 | 61000 | 19973 | |
| m5.2xlarge | 3000 | 41000 | 42000 | 18375 | |
| m5.xlarge | 30000 | 67000 | 90000 | 21386 | |
| m5.xlarge | |||||
| m5.xlarge | |||||
| m5.xlarge | 30000 | 76000 | 121000 | 21975 | |
| 4.15 | m5.xlarge | 8000 | 41000 | 61000 | 19593 |
| m5.xlarge | 9000 | 41000 | 61000 | 19545 | |
| m5.xlarge | 8000 | 41000 | 51000 | 19500 |
Expected results:
4.16 podLatency should match 4.15 results
Additional info:
Investigation under progress.
