manticoresoftware
rejoin cluster after alter cause joiner to exit with the error
if one node in the cluster was shut for a period of time and cluster adds index then the old node that joins the cluster back and receives IST fails to apply ALTER CLUSTER name ADD index as it has not local index and it exit the cluster with the error.
For that cause it could be better to joiner node to join the cluster via SST but that needs that all other nodes to keep the latest ALTER seqno and joiner node should check for the newest cluster ALTER seqno and issue join from the scratch instead of join via IST.
To force SST on the joiner node daemon could set seqno at the grastate.dat to -1 prior calling the Galera code.
the issue was for a long time but poped up during the fix of https://github.com/manticoresoftware/manticoresearch/issues/1802