Replication Link Failure
Hitachi arrays send an alert in the following situations:
- When a replication link failure is detected
- When any P-VOLs, on which data has been written, transition from the PAIR state to the PSUE state.
In fence levels never and async, a replication link failure does not compromise the application's ability to write to its local devices; the arrays start tracking changed regions on disk in preparation for resynchronization when the link is restored.
The devices do not automatically resynchronize when the link is restored, nor do they change state once the restoration is detected. An administrator can resynchronize the devices, either from the command line or by running a configured action using the agent's action entry point. The following situations require administrative action after a link failure is repaired. These actions depend on the fence level and any events that may have occurred during the failure.
| Event
|
Fence Level
|
Recommended Action
|
|---|
Link fails and is restored, but application does not fail over
|
never, async
|
Run the pairresync action to resynchronize the S-VOLs
|
Link fails and application fails to the S-VOL side
|
never, async, or data
|
Run the pairresync-swaps action to promote the S-VOLs to P-VOLs and resynchronize the original P-VOLs.
|
Application faults due to I/O errors
|
data
|
Run the localtakeover action to write-enable the local devices. Clear faults and restart service group.
|
|
