Lukasz Sojka

> I agree that the nemesis shouldn't fail if it was able to delete at least one. What happened after the deletion? Was there an exception? the nemesis didn't continue...

from events: ``` 2021-12-16 18:09:58.477: (DisruptionEvent Severity.ERROR) period_type=end event_id=02fdbdae-a39d-4989-8ea4-027e7d56fd6c duration=2m52s: nemesis_name=DestroyDataThenRebuild target_node=Node longevity-large-partitions-3h-4-6-db-node-8059573b-5 [52.209.60.29 | 10.0.0.170] (seed: False) errors=Data file for destroy is not found in scylla_bench/test Traceback (most recent...

We didn't run this nemesis in this test for 5.0/2022.1

> @soyacz for now, let's change the seed of this test and run it with different set of nemesis. Started test with different seed: https://jenkins.scylladb.com/job/enterprise-2022.2/job/scale/job/scale-5000-tables-test/5/

> > > @soyacz for now, let's change the seed of this test and run it with different set of nemesis. > > > > > > Started test with...

why do we need warmup in Gemini?

we got reproduction with s-b `0.1.3` executed from docker. Also I see from commit history, that on 14 January 2022 we bumped up version of go. Maybe this is rather...

@fruch shall we try building scylla-bench image with base go image version 1.16 and see if it reproduces?

I'm not sure, disconnections were persisting sometimes for 2 minutes. We would need to test it.

I tried with timeout settings like this: ` -timeout 15s -retry-interval=80ms,5s -retry-number=20` and it failed anyway.