Botond Dénes
Botond Dénes
New in v3: * Rebased.. * Reorder pause_guard and cpu_need_guard to avoid permit in `active_await` state being paused (this caused the above CI failures).
> [test_tablet_drain_failure_during_decommission.dev.5](https://jenkins.scylladb.com//job/scylla-master/job/scylla-ci/12396/testReport/junit/%28root%29/test_topology_failure_recovery/Tests___Unit_Tests___test_tablet_drain_failure_during_decommission_dev_5) [🔍](https://github.com/scylladb/scylladb/issues?q=is:issue+is:open+test_tablet_drain_failure_during_decommission.dev.5) https://github.com/scylladb/scylladb/issues/20060
> ❌ - [Unit Tests](https://jenkins.scylladb.com/job/scylla-master/job/scylla-ci/12406/) Infra issue.
> > ❌ - [Unit Tests](https://jenkins.scylladb.com/job/scylla-master/job/scylla-ci/12406/) > > Infra issue. And also https://github.com/scylladb/scylladb/issues/19969.
New in v3: read at least a full buffer, even if hint allows stopping earlier.
New in v4: * Include https://github.com/scylladb/scylladb/pull/20725 * Add unit test * Conclude performance testing
New in v5: don't copy mutations across shards in the test (use frozen mutation instead).
Dup of https://github.com/scylladb/scylla-dtest/issues/5866.
The fix was merged in https://github.com/scylladb/scylla-ccm/pull/676, closing.
Just found out we have the same problem in https://github.com/scylladb/scylladb/blob/0114e4c2ae11bdcc6c7348692bef61d357403b6c/service/storage_proxy.cc#L3827 I don't understand how we are not hitting this more. Is this related to the boost::ranges->std::ranges migration? Is boost eager...