During mirroring of data between two clusters, the MM2 connector / task failed with an NPE but for Connect is kept reporting s running. This is an issue because it is not obviously clear that something is broken and only later the user finds that mirroring stopped. The connector needs to change into the failed state in case like this.

However - despite it sounding like a nasty bug - right now it is not completely clear what actual cause / issue is. It might be complicated to fix / it might take time to fix.