Using External Snapshot Functions for Disks and Devices Protected by DRBD

An external snapshot can be used to restore data for a DRBD volume. LifeKeeper requires “internal” metadata on all DRBD resources to keep the actual data and metadata “inextricably” linked by storing the metadata and the actual data on the same device. When a snapshot is used to restore data, DRBD is able to determine the state of the actual data from the metadata to allow for partial resynchronization and to assure consistent data. A partial resynchronization will automatically occur from the server with existing data to the server that was restored from a snapshot.

Snapshot capability referred to in this document includes:

1. Snapshots provided by cloud environment services such as AWS
2. Snapshots provided by virtualization software such as vSphere
3. Snapshots provided by shared storage in physical environment

The recommended procedure for using snapshots is as follows:

1. Bring the DRBD resource/hierarchy in-service on the server that has good data, i.e. the server that is NOT going to have the volume from the restored snapshot.
2. Stop LifeKeeper (lkstop) on the server where the volume will be attached from the restored snapshot.
3. Detach the old volume.
   1. Record the location the volume is connected (e.g., /dev/sdb).
   2. If the OS supports “hot add”, then the device nodes for the detached volume should automatically be removed when the volume is detached. If they are not then shutdown the OS (shutdown –poweroff).
4. Create the new volume from the snapshot
5. Attach the new volume on the same location recorded in step 3a as the old volume that was deleted.
   1. Restart the OS if it was shutdown in step 3b.
6. Verify that the new volume is accessible on the same device as the old volume.
7. Start LifeKeeper (lkstart).
8. The server that has the DRBD resource/hierarchy in-service will automatically connect the DRBD resources and start resynchronization.