We had some strange problems during the migration of our three-node QUAL cluster.
During the migration, the file /etc/my.cnf was saved into /etc/my.cnf.rpmsave and a new one was created containing only two lines :
The first directory was empty, the second one contained two file : mysqld.cnf almost empty and the second one wsrep.cnf almost empty also and containing the line
After the upgrade of the three nodes, everything seemed to be working well. There was no alert in ClusterControl but in fact, we had lost the configuration parameters of the database and also for the replication. We had to put them back in the two new files to make everything running properly.
Almost one week later, we had to reboot the database nodes for maintenance and we noticed another problem:
- the file /etc/xinetd.d/mysqlchk was also saved at the time of the upgrade to mysqlchk.rpmsave and the new file contained the line "server /usr/bin/clustercheck" insted of previously "server /usr/local/sbin/mysqlchk".
The first command /usr/bin/clustercheck gives an HTTP 503 error whilst the second one runs fine.
We had to modify the /etc/xinetd.d/mysqlchk on the three database servers in order to be able to access the database through the HAProxy nodes and the VIP (Keepalived).
I wonder if somebody else has already experienced these errors. Note that our local partner had tested the procedure using Centos servers and they didn't have all these problems.
Please sign in to leave a comment.