Full synchronization is required when using a snapshot process to restore data to a disk or device that is actively protected by DataKeeper.
Snapshot capability referred to in this document includes:
- Snapshots provided by cloud environment services such as AWS
- Snapshots provided by virtualization software such as vSphere
- Snapshots provided by shared storage in physical environment
The process of restoring snapshots typically takes place without involvement of the operating system. When the snapshot is restored without involving the OS DataKeeper cannot properly synchronize these changes to the target system. In addition, the changes to the underlying disk or device are not written in the bitmap, so consistency is not maintained with differential synchronization.
The recommended procedure for using snapshots is as follows:
- Stop the mirror with “pause mirror” command
- Restore the snapshot
- Perform a complete re-synchronization with restored data as the source
The source and target data are in an inconsistent state until the mirror is fully synchronized. Therefore, it is essential to execute full synchronization after the snapshot has been restored. Please note a full resync can take a long time depending on the mirror size, available bandwidth, and system resources. During the resync period, the DataKeeper resource and any dependent resources cannot be switched over.
Full synchronization may not be required when restoring the same snapshot to both the source and the target, but this operation is not supported.
Post your comment on this topic.