When upgrading from 7.0.2 to 7.1.1, it is observed that a watcher background thread was created by metadata_provider to connect to other indexing nodes which never got cleaned up. this thread had stale credentials and floods the logs with connection errors. need to investigate why such a thread was never killed.
|For Gerrit Dashboard: MB-55512|
|186490,3||MB-55512 kill orphan metadata provider threads in rebalanceDone||neo||indexing||Status: MERGED||+2||+1|