Rollback can be done only from the backups that were enabled for Instant Recovery and made with one of the following methods: VxFS_Checkpoint, VxVM, FlashSnap, NAS_Snapshot, or the disk array methods.

If the backup was made with the EMC_TimeFinder_Clone method and the clone is not fully created, a rollback cannot succeed.

To verify that the clone is complete before you do a rollback:

See Verifying that the clone is complete before doing a point in time rollback.

For the backups that were made with the VxFS_Checkpoint method, rollback requires the VxFS File System 4.0 or later and Disk Layout 6. For NAS_Snapshot, the file system requirements depend on the NAS vendor.

Rollback deletes any VxFS_Checkpoint snapshots or NAS_Snapshot snapshots (and their catalog information) that were created after the creation-date of the snapshot that you restore.

If the primary file system is mounted and the snapshot resides on a disk array, the rollback attempts to unmount the file system. Any I/O on the primary device is forcibly stopped if the unmount succeeds. To be safe, make sure that no I/O occurs on the primary device before a rollback.

If the attempt to unmount the primary file system fails, the rollback does not succeed. You should halt I/O on the device and retry the rollback. For example, if a terminal session has accessed the file system through the cd command, change to a directory outside the file system and retry the rollback.

Rollback is available only when you restore the file system or volume to the original location on the client.

When a file system rollback starts, NetBackup verifies that the primary file system has no files that were created after the snapshot was made. Otherwise, the rollback aborts.

Rollback from OST_FIM type snapshot can be done from copy one only.

For the rollback from OST_FIM type snapshot, refer to the NetBackup Replication Director Solutions Guide.