How it Works

When the maintenance mode flag is set to true, the cluster role changes and GR is triggered for the rack. The standby cluster takes over the responsibility of the cluster that is in the maintainence mode. During this period, the monitoring threads check the runtime value of the flag and pause the execution when the maintenance mode flag is set to true. By default, for fresh installation, the flag is set to false.

Both, the source and standby clusters (racks) can be under the maintenance mode at the same time. You can enable the maintenance mode for the rack server regardless of its state.

You can push the system to the maintenance mode when the maintenance procedure is in-progress for the mated cluster. Before you start the maintenance activity, set the geo maintenance mode flag value to true. When the maintenance is complete, reset the flag to false after confirming the health of the system.

For information on how to configure the maintenance mode flag, see Enabling or Disabling Maintenance Mode.

When the maintenance mode is enabled:

  • Automated GR-switchover such as pod monitoring, BFD link monitoring from the rack server is not supported.

  • Only CLI-based GR-switchover is supported from the rack (with maintenance mode enabled) to the partner rack.

  • GR-switchover, including CLI-based, is not supported from the partner rack to the rack where the maintenance mode is enabled.

  • If both partner racks are in the maintenance mode, GR-switchover is not supported.

  • All the monitoring activities are paused.

  • The mated cluster cannot trigger the failover when it detects the local failure.

  • Replication activities continue on the cluster.

  • Maintenance mode doesn't implicitly change the instance roles of the site. However, role change is possible using the geo switch-role role CLI command.

    GR trigger is not supported towards and from the cluster that is under maintenance. Only CLI-based failover is supported from the cluster under the maintenance.