Kamil Braun

`node1 is still running after 127 seconds. Trying to generate coredump using kill(25663, SIGQUIT)`

We need to fix https://github.com/scylladb/scylladb/issues/16604 which will make such failures more obvious.

Anyway, @temichus @aleksbykov we need to modify this test, in raft-topology mode this part: ```python3 logger.debug(f"Restart the cluster with shards {smp_after}") for node in self.cluster.nodelist(): logger.debug("Starting node " + node.name)...

moving from required to optional list for raft-topology project - test not in gating - expected behavior, test needs to be adjusted instead of scylla

`test_hintedhandoff_rebalance` needs to be adjusted the same way

There are two remaining dtests suffering from this that need to be adjusted: https://github.com/scylladb/scylla-dtest/issues/4068 https://github.com/scylladb/scylla-dtest/issues/4070

Looks like we have some kind of performance regression on read path when running with raft topology? We should definitely not mask it using increased timeout but dig and understand.

@temichus can you verify the test for flakiness?

@temichus it looks like the run used outdated Scylla version ``` [2024-03-19T11:02:15.302Z] docker-image-name-aarch64: scylla-nightly:5.5.0-dev-0.20240313.15bea069a911-aarch64 [2024-03-19T11:02:15.302Z] docker-image-name-x86_64: scylla-nightly:5.5.0-dev-0.20240313.15bea069a911-x86_64 [2024-03-19T11:02:15.302Z] [email protected]:scylladb/scylla.git-sha: 15bea069a911a9c4e5dee23beae5029819de7224 ``` 15bea069a911a9c4e5dee23beae5029819de7224 does not include the fix 5ef47c42b3639c8490c3b5dbc171abd00f44801c

Nice catch! Isn't this the likely cause of https://github.com/scylladb/scylladb/issues/15924?