[xCluster] Race condition with SPLIT_OP processing and CdcConsumer shutdown #12068
Labels
area/docdb
YugabyteDB core features
kind/bug
This issue is a bug
priority/medium
Medium priority issue
Projects
Jira Link: DB-836
Description
Noticed when working on
XClusterAutomaticTabletSplitITest.AutomaticTabletSplitting
.There is a race occurring during the shutdown path, if we shutdown immediately after sending a DeleteUniverseReplication request and before waiting for all the CDCConsumers to be properly cleared.
From my investigations, we seem to be getting stuck in
rpcs_->Abort({&write_handle_});
during~TwoDCOutputClient()
, but unclear as to what its waiting on..Currently working around in tests by waiting for pollers to be deleted by DeleteUniverseReplication before continuing with the shutdown.
The text was updated successfully, but these errors were encountered: