old slave session after cluster node failure

[jdanilson]

We are running replication_manager.sh between two pxc clusters of three nodes each. Operation is flawless so far. But we note, upon failure of a slave node, even a planned shutdown, there may be (not always) residual slave artifacts left on the failed node upon restart.

When the failure or shutdown occurs, we see the slave sucessfully moved to a new node. No issues there. But upon restart we may see the old slave still active either with one or both threads active. This zombie slave appears not to do damage (likely due to the gtid usage) but confuses our monitoring.

Two questions:

1. is this expected?
2. what is the best method to clean the zombie slave up? Thinking reset slave all but perhaps there is a better way.