Details
-
Bug
-
Resolution: Unresolved
-
Major
-
6.6.0, Cheshire-Cat
-
Untriaged
-
1
-
Unknown
Description
Inspired by this bug: MB-45064
DCP replicator might get stuck while opening/closing the stream on the node that does not respond. If this happens during the rebalance, the rebalance hangs bu the replicators will be nuked as soon as the rebalance is aborted.
If this happens outside the rebalance (janitor, failover) there's no mechanism that allows to nuke stuck replicators. It looks like it won't even be possible to do failover the dead node due to janitor_agent being stuck.
There's no known CBSE so far where this happens.