Kamil Braun

This always happens during "second node rollback" stage. We start background reads, rollback the node, then validate the background reads, and timeout. Curiously, the test does a similar thing earlier,...

I sent a PR to disable this read stress part of the test so we can keep the test in gating: https://github.com/scylladb/scylla-dtest/pull/4195 It should still be investigated.

I'm taking this into my backlog.

Removing P1, the failure was happening since forever, and that part of the test was disabled for now so we can keep it in gating.

> Seen in unit tests too: https://github.com/scylladb/scylladb/issues/17353 Different symptoms. In the unit test we're getting a query failure (driver returns NoHostAvailable). Here we see cassandra-stress hanging.

This test doesn't use the Python driver, the problem is with a timing out cassandra-stress command, which uses Java driver underneath. (But we don't know if the problem is with...

6.0 is unrealistic. We have greater priorities.

From @michoecho's investigation, it's caused by `apply_fence`. ` seastar::continuation::run_and_dispose `  Citing some comments from the discussion: > (If this wasn't clear, note that the "bad patch" doesn't actually add...

@avikivity do you consider this a release blocker?

I put it in Q2 plan anyway, we might be able to fix it for 6.0 even after we branch, if we deal with all the blockers.