Note
There may be preparatory steps that you would like to take before performing failover or while your cluster is operating normally in order to be prepared to switch client applications over to the replication peer. See Deploying a Failed Over Replication Peer as a Working Cluster.
If you are not sure which type of failover to perform, see Failover Capabilities and Scenarios.
This option is available if the peers are connected.
On the destination peer, open the Data Protection page and select the Protected Paths tab.
Identify the protected path configured for native replication. The state should be "Active" and the role should be "Destination".
-
Click
to open the Actions menu for the protected path and select Modify Replication State. -
Verify that Fail over gracefully is enabled and review the current and result states shown in the dialog:
The current state shows replication from the writeable source peer to the read only destination peer. The result state shows a reversal of the replication direction and the writeable and read only states:
Tip
The following warning may appear:
Configured replication interval is X. Not meeting RPO - replication interval is off by Y. Estimated Read-Only time is (X+Y)*2.
This message warns you that the downtime during which the data paths on both peers will be read-only is longer than expected. In this context, RPO is said to be met if replication is completed without the snapshot schedule being disrupted. If replication is not complete when a snapshot is scheduled, the snapshot is skipped.
Assuming you are satisfied that the result state shows the desired result, click Fail Over.
-
Click Yes to confirm the failover.
The failover starts immediately, with the source peer becoming read only, and proceeds as described in Ungraceful Failover.
You may see the Role value in the Protected Paths tab change states during the process. When the process is done, the role is displayed as Source and the protected path state active.
Warning
In ungraceful failover, the following data are lost: Any data that was written to the source peer since the point in time when the last successfully replicated snapshot was taken.
This type of failover is available even if the peers are not connected. The lack of connection is indicated by the State of the replication peer configuration as shown in the Replication Peers tab of the Data Protection page. When the peers are not connected, the state is reported as Not Connected.
On the destination peer, open the Data Protection Page and select the Protected Paths tab.
-
Click
to open the Actions menu for the protected path and select Modify Replication State. -
Disable the Fail over gracefully setting.
The dialog shows that in the current state, the source peer is writeable and is replicating to the destination peer which is read-only. The result state shows that if you proceed with failing over, replication will be suspended and both peers will become writeable.
-
Click Fail Over and then click Yes to confirm the action.
You may see the Role field in the Protected Paths tab reflect changes during the process. When the process is done, the role is displayed as Standalone.
The resulting State of the protected path is either of the following:
Active, if the protection policy states to keep local snapshots.
Suspended, if the protection policy has no retention time for local snapshots.
To fail back to the primary cluster after a graceful failover, repeat the graceful failover procedure in the opposite direction, starting on the current destination peer.
After an ungraceful failover, if and when the connection between the peers is re-established, a failback procedure can be performed.
This procedure enables you to resume working with the primary cluster and will sync that cluster with the data that was written after the fail over to the secondary cluster while the peers were disconnected. Any data that was written to the primary cluster since the last snapshot that was fully replicated to the secondary cluster prior to the ungraceful failover is discarded.
-
Click
to open the Actions menu for the protected path and select Modify Replication State.The dialog shows the current state in which the two peers are both in standalone role, both writeable with replication suspended.
The result state is that the current local cluster from which you're initiating this action will remain writeable, become the source peer and begin replicating to the other peer which will become read-only. The other peer will become the destination peer.
-
Click Fail Over and then Yes to confirm the action.
You may see the Role field in the Protected Paths tab reflect changes during the process. When the process is done, the role is displayed as Source.
-
Wait until a restore point is fully replicated on the destination.
-
On the destination peer, click
to open the Actions menu for the protected path and select Modify Replication State.You will now have the option to perform a graceful failover.
Leave the Fail over gracefully option enabled.
-
Click Fail Over and then Yes to confirm the action.
You may see the Role value in the Protected Paths tab change states during the process. When the process is done, the role is displayed as Source and the protected path state still active.
Comments
0 comments
Article is closed for comments.