MariaDB 10.1 Galera cluster - new changes in setup
Was doing a POC on Galera cluster on MariaDB 10.1 and below are some of the points where we had to troubleshoot when setting up based on below Mariadb document --
https://mariadb.com/resources/blog/setting-mariadb-enterprise-cluster-part-1-how-set-virtual-machine-template
Above document is based on MariaDB 5.5 hence the below changes were required.
1. From new version "wsrep_on=ON" parameter must be set.
In all cluster nodes /etc/my.cnf.d/server.cnf should be edited to set wsrep_on=ON. Other wise wsrep_ready parameter will be in off status and cluster will not be functional.
2. When bootstraping the cluster first node must be started with "galera_new_cluster" command. mysqld --wsrep-new-cluster is not supported in new version.
3. Both SElinux and firewalld needs to configured (Stopped) to allow required communication between cluster nodes.
If there any "[ERROR] WSREP: wsrep::connect(gcomm:" errors recorded in the logs that can be due to the blocked communication from firewall or SElinux
https://mariadb.com/resources/blog/setting-mariadb-enterprise-cluster-part-1-how-set-virtual-machine-template
Above document is based on MariaDB 5.5 hence the below changes were required.
1. From new version "wsrep_on=ON" parameter must be set.
In all cluster nodes /etc/my.cnf.d/server.cnf should be edited to set wsrep_on=ON. Other wise wsrep_ready parameter will be in off status and cluster will not be functional.
2. When bootstraping the cluster first node must be started with "galera_new_cluster" command. mysqld --wsrep-new-cluster is not supported in new version.
3. Both SElinux and firewalld needs to configured (Stopped) to allow required communication between cluster nodes.
If there any "[ERROR] WSREP: wsrep::connect(gcomm:" errors recorded in the logs that can be due to the blocked communication from firewall or SElinux
Comments
Post a Comment