I have been using clustercontrol CE for a couple of month while testing our galera-cluster. While testing we found an issue with all the writes going to all nodes in our cluster - even we had max-writers = 1 in our ProxySQL hostgroup configuration.
After spending some time testing different hostgroup-configurations - which did not solve the issue - we (instead of relying on clustercontrol's clusterload 'with Show servers clicked') digged into proxy discovered that we did NOT have any issues, and the writes did indeed go to the designated writer-node. Even clustercontrol showed that writes were distributed on all the nodes.
I really like clustercontrol, so i need to ask - Why does it looks like write-queries are distributed to all nodes when its not? Did i miss something or did i set it up the wrong way?
Please sign in to leave a comment.