The 'clustercheck' helper script should read all variables from /etc/sysconfig/clustercheck. This file contains information required to check mariadb-galera cluster state, such as username, password, port, etc. If clustercheck cannot read variables from /etc/sysconfig/culstercheck, it will use default values which will almost certainly fail. Running clustercheck on RHOS-5.0, RHEL-6.5 currently gives: # clustercheck HTTP/1.1 503 Service Unavailable Content-Type: text/plain Connection: close Content-Length: 44 Percona XtraDB Cluster Node is not synced. In this particular case this is wrong. The node is an synced member of the mariadb-galera cluster. The expected result is: # clustercheck HTTP/1.1 200 OK Content-Type: text/plain Connection: close Content-Length: 40 Percona XtraDB Cluster Node is synced. Using 'bash -x' to run clustercheck makes the undelying mysql command visible, and clearly shows that the default values are being used, not the values in /etc/sysconfig/cluster.
I actually fixed this upstream a while back: https://github.com/rohara/percona-clustercheck/commit/d0f366eeb91b72aaf022342b886062366a4e5f63 But it never got pulled into a RHOS-5.0 build. Easy fix, just requires rebuild.
The clustercheck script in rhos-5.0-rhel-7 branch is already up to date. This bug only exists in rhos-5.0-rhel-6 branch.
part of an HA setup