Check Galera cluster leader
To stop or start the Galera cluster you should first identify the leader node.
On the datanodes, check the state of Galera:
[root@datanodes0 /]# cd /var/lib/mysql/ [root@datanodes0 mysql]# cat grastate.dat # GALERA saved state version: 2.1 uuid: c0e2c5f4-e045-11e9-988e-d6d16254ad99 seqno: -1 safe_to_bootstrap: 0
If there is a node with safe_to_bootstrap=1, you should start the cluster on this node
If no node is marked as safe_to_bootstrap, force it (using the virt IP).
Stop Galera cluster
To stop the Galera cluster,
- Find the leader node as described above
Starting with the nodes that are not the leader. Stop the database on all nodes, .
[root@datanodes2 /]# systemctl stop mariadb [root@datanodes1 /]# systemctl stop mariadb [root@datanodes0 /]# systemctl stop mariadb
Start Galera cluster
To start the Galera cluster, log in to the leader node.
Start the cluster with this command
[root@datanodes0 /]# galera_new_cluster
On the other nodes, just start MariaDB or MySQL service
[root@datanodes0 /]# systemctl start mariadb
Helpful commands to check cluster status:
[root@datanodes1 mysql]# maxadmin show servers [root@datanodes1 mysql]# maxadmin list servers
RabbitMQ Cluster check and rejoin
On each node do these commands.
rabbitmqctl stop_app rabbitmqctl join_cluster rabbit@datanodes1 rabbitmqctl join_cluster rabbit@datanodes2 rabbitmqctl start_app rabbitmqctl cluster_status