PowerStore: Increased physical capacity due to stale replication snapshots

Appliance physical capacity may grow larger due to stale replication snapshots.

This issue affects PowerStore asynchronous replication.

PowerStore asynchronous replication is based on a predefined Recovery Point Objective (RPO). The RPO defines (in minutes) the time difference between each of the replication updates.

PowerStore has two types of RPO-based replication: NAS and Block replication

Following the initial replication, PowerStore marks the baseline as T0 and after a specified RPO creates an RPO snapshot (T1). The RPO snapshot is used to send the differences between the baseline snapshot (T0) and the RPO snapshot (T1).

Once the delta between T0 and T1 is replicated, PowerStore marks the new RPO snapshot (T1) as the new baseline and expires the old RPO snapshot taken at T0. After another specified RPO, another snapshot is created (T2) and once the changes are replicated, it is promoted as the new baseline and expires the old RPO snap taken at T1.

This flow repeats endlessly.

If during the phase where PowerStore promotes the new baseline snapshot, the management network is disconnected for any reason, there is a chance the new RPO snapshot is not promoted as the new baseline snapshot and become a stale snapshot, thereby consuming space. The next time a new RPO snapshot is created, once replication is done, that RPO snapshot will be used as the new baseline, but the old RPO snapshot will remain in the system indefinitely.

Resolution

The issue is fixed in PowerStoreOS version 4.0.0 or later.

Workaround

Contact Dell Technical Support to help with a workaround. Reference this KB Article in the Service Request for expedited handling. 

Once the workaround is applied, schedule an upgrade to PowerStoreOS 4.0.0 or later.