When an old primary cluster is restarting for the first time after a successful takeover, the MySQL database does not detect that the cluster should no longer act as a master and the Geographic Edition software still keeps the primary role, but leaves it deactivated. The goal for the recovery is to configure the old master to run as a slave and to update the Geographic Edition software configuration to reflect this role change.
You can check for the status with the following command:
# geoadm status
The recovery strategy after a takeover involves the following actions:
Configuring the old master to run as a slave
Manually starting the slave threads on the old master
Resynchronizing the protection group to switch the role
Remove the skip-slave-start keyword from the appropriate my.cnf file.
mysql> start slave;
mysql> show slave status\G
If the slave status shows that at least one slave thread is not running, fix the root cause, and retry the operation. As a last resort, you could take a backup from the current master and perform a fresh slave setup.
Connect to a node of the old primary cluster and update the protection group to change the role from a deactivated primary cluster to a secondary cluster.
You must be assigned the Geo Management RBAC rights profile to complete this procedure. For more information, see Securing Geographic Edition Software in Oracle Solaris Cluster Geographic Edition Installation and Configuration Guide .
# geopg update protection-group
# geopg start -scope local protection-group
For more information, see Resynchronizing a Protection Group in Oracle Solaris Cluster Geographic Edition System Administration Guide and Activating and Deactivating a Protection Group in Oracle Solaris Cluster Geographic Edition System Administration Guide .