Artsiom Mishuta

Results 103 comments of Artsiom Mishuta

Test read has a big latency in **"Total test HDR Histogram percentiles"**, but as I mentioned in https://github.com/scylladb/scylla-cluster-tests/pull/8052. Hints are not creating during **the read workload** Test: performance_regression_test.PerformanceRegressionTest.test_latency_read_with_nemesis Test start...

p99>10 ms in 1 cycle in add_new_nodes in **READ** workload (Hints are not creating during the read workload) Test: performance_regression_test.PerformanceRegressionTest.test_latency_mixed_with_nemesis Test start time: 2024-07-18 15:02:42 Test id: e4fae0f7-dbce-4db1-9466-06cc6ccd4f78 Scylla Server...

fresh bisect runs for hinted-nandoff-latency-650gb-with-nemesis and branch-perf-v15 branches mixed hinted-handoff 2fb3c7c9-880a-41a3-a4cf-c4695546b609 (in add_new_nodes P99 > 10 ms) ![image](https://github.com/user-attachments/assets/5c31aee4-57c7-4cc8-ba25-69b64d576ab6) ![image](https://github.com/user-attachments/assets/5961ec48-9bcf-4a10-9c4d-521504904e6d) ![image](https://github.com/user-attachments/assets/8f07e2f4-8e41-4201-b3c0-253ea85c78ac) ![image](https://github.com/user-attachments/assets/60749fa3-674e-4616-bbde-0829ea699291) ![image](https://github.com/user-attachments/assets/246656a2-d9a0-4e78-9110-b4a99e670797) mixed branch-perf-v15 9b02948e-d859-4745-9963-85a4d1fb500c (in add_new_nodes and _mgmt_repair_cli P99...

read hinted-handoff 3a07b87b-c496-4415-a9f7-f8cf4ff71154 (only Total test HDR Histogram percentiles RED but nemeses are good) ![image](https://github.com/user-attachments/assets/9129a0d3-1a0b-4342-ab3b-cdd445714960) ![image](https://github.com/user-attachments/assets/90fbcee9-fd21-4539-a25f-246bdaa19c7c) ![image](https://github.com/user-attachments/assets/cc0d3762-fe45-48b3-b439-ab3273091459) ![image](https://github.com/user-attachments/assets/9b711e80-0d07-4986-993d-0ef0dd2667c1) ![image](https://github.com/user-attachments/assets/155eb8b7-dd2f-427b-9c50-39b129009660) ![image](https://github.com/user-attachments/assets/ae5aea6f-7d60-45bf-bc52-54ffe722d1aa) read branch-perf-v15 ee9fc26a-600c-4424-8946-483029179300 ![image](https://github.com/user-attachments/assets/d0890f4a-651b-45c3-8790-6f092b7678e3) ![image](https://github.com/user-attachments/assets/bc0c80e4-9fe8-4b45-b7b7-1b4a68e03746) ![image](https://github.com/user-attachments/assets/48554d99-ef38-4fd5-a56a-33e96e3f0b4d) ![image](https://github.com/user-attachments/assets/0b86ff34-53a8-48d3-9149-f6426bcd39d1) ![image](https://github.com/user-attachments/assets/b71cf5bd-854a-48b3-907e-147803951445)

write hinted-handoff 36e3aad7-146f-463d-80a8-b233e75ab0eb ![image](https://github.com/user-attachments/assets/b59a6e9f-b74f-4d8e-aad7-7d8291234d0f) ![image](https://github.com/user-attachments/assets/2743fbbf-e3d6-40d2-88a6-068d6d41b38f) ![image](https://github.com/user-attachments/assets/5fdb36c9-d8ce-478e-8527-c31f8c44de4a) ![image](https://github.com/user-attachments/assets/a5475247-dda5-45e8-bf66-d6f6a7a313a9) ![image](https://github.com/user-attachments/assets/b1d59e71-1521-4cef-8c44-137d8adb44bf) write branch-perf-v15 1d0af835-a5fa-4eea-b7cc-bd579b472e56 ![image](https://github.com/user-attachments/assets/514a1797-cafc-4673-a57b-7b2239f9a2eb) ![image](https://github.com/user-attachments/assets/e6b0ca65-d85b-4c32-b81f-3a3ed79d3a83) ![image](https://github.com/user-attachments/assets/1d31b31c-faee-42c8-bceb-2afaf39fc523) ![image](https://github.com/user-attachments/assets/db230120-b3fa-4136-8546-975f19962c6c) ![image](https://github.com/user-attachments/assets/aa5080f1-6db0-48d0-9389-d4e5cb85c03f)

@roydahan, @juliayakovlev according to https://github.com/scylladb/qa-tasks/issues/1705#issuecomment-2491220067. fresh bisect runs for hinted-nandoff-latency-650gb-with-nemesis and branch-perf-v15 branches showed almost the same results. the difference is only in **READ** workload: hinted-handoff run shows **RED P99**...

> Do we want this test will always run with HH enabled? Not new additional test? yes, we want to run perf tests with HH enabled. because it is enabled...

@fruch, @juliayakovlev, @roydahan PR rebased/ merge conflicts solved. Will you accept this PR to merge?

this > > last time seen in SCT tier1 2024-06-29 > > @temichus , can I see logs from this run? If it's on fresh enough scylla version, it should...