Is it correct to assume that maxscale periodically checks the 'mysql' db for monitoring purposes?
On large imports it seems to take too long to connect (or the table is locked) and the check runs into a timeout.
This leads to following action:
every node that cant be checked gets removed from the cluster.
Log: Readwritesplit service can't locate the master. Client sessions will be closed.
The only way to get around it is to up the timeout values of the instances monitor settings.
Is this intended behaviour? Should we do something other than tweaking the timeout values?