If the primary cluster becomes unavailable, you can perform a manual failover by promoting the secondary cluster to primary. This operation will switch the configuration of the secondary cluster, including its DNS name, to the configuration of the primary one. Failover of the primary cluster can be performed in the following cases:
- The current primary cluster is completely non-operational and isolated from the Internet and any backup agents.
- Backup agents are unable to communicate with the current primary cluster.
- The DNS name of the primary cluster has been reconfigured to its IP addresses.
Promoting the secondary cluster to primary is an irreversible operation that will invalidate all data on the primary cluster. Use it only in case of emergency.
- Geo-replication is enabled, as described in Enabling geo-replication.
- The secondary cluster can reach itself via its own domain name on TCP port 44445.
To perform a failover
- On the secondary cluster, go to Storage services > Backup storage > Geo-replication, and then click Promote to primary.
- Click Failover in the confirmation window.
If the current primary cluster is still operational, forcibly release all of its nodes from Backup Gateway first and then perform a failover.
Run the following command:
# vinfra service backup geo-replication slave promote-to-master