Closed aleksbykov closed 1 year ago
There are just 2 stalls, and they look like flukes. The performance degradation is likely unrelated.
Nothing in the metrics indicates a problem.
@avikivity Should i close the issue?
Is there a regression or not?
The total max throghput is less by 11%, the only problem is reactor stall but you said it is flukes and metrics shows nothing. Here is results:
Current test result | Version compared to | Best | Diff best | Commit, Date | Last | Diff last | Commit, Date |
---|---|---|---|---|---|---|---|
89336.0 | 5.3.0~dev | 100631.0 | -11% | #941407b905a62023-02-18 | 91607.0 | -2% | #3b424e391b6c2023-05-21 |
89336.0 | 5.4.0~dev | 99022.0 | -9% | #b7022cd74e7a2023-06-13 | 78563.0 | +13% | #f6f974cdeb112023-06-29 |
It looks like regression, but may it is fluky results related to aws
@avikivity could you also take a look on that stalls, the happened for same build but during latency mixed test. decoded-reactor-stalls.tar.gz
@avikivity could you also take a look on that stalls, the happened for same build but during latency mixed test. decoded-reactor-stalls.tar.gz
These look like #8828
8828
Thank you
Version: 5.4.0~dev
Performance throughput test for master shows throughput degradation on 10% for last several runs. During last runs reactors stalls up to 34ms were registered on 2nd node.
DB logs: https://cloudius-jenkins-test.s3.amazonaws.com/a4ef6573-9210-4176-974c-7fa13d4c0b73/20230709_052919/db-cluster-a4ef6573.tar.gz
Monitoring: http://44.204.125.133:3000/d/advanced-master/advanced?orgId=1&from=1688876106352&to=1688879512229