Skip to main content

Our CommVault instance uses Nimble storage for data backup. When CommVault tries to backup VM, it tells vCenter to take a snapshot of this VM. After snapshot is created by vCenter, CommVault backup it.

At the same time we have Nimble replication process. When Nimble discovers a new snapshot (created by CommVault), it puts the snapshot in the queue for the replication.

When CommVault finishes backup, it tells vCenter to delete snapshot. But if the replication of the snapshot was not finished, Nimble replication process prevents snapshot being deleted. At the same time vCenter does not generate any failure for the CommVault.

At the very end:

-          CommVault does not repeat an attempt to delete because it does not know that this attempt failed (I tried to decrease nExpireSnapshotHours parameter value, but this do not help; other snapshot-related parameters are not applicable because no failures nor error are generated);

-          vCenter seems to lose the control/ownership on this snapshot and does left this snapshot as "being online";

-          Nimble does not delete "a foreign" snapshot too.

And the count of snapshot becomes invincible - till it becomes greater than the maximum count that Nimble allows to maintain.

 

The question is:

-          does CommVault have a special parameter that allow to increase the interval between the end of backup time and the time when CommVault asks vCenter to delete the snapshot?

I've checked CommVault Additional settings, but found no parameter that defines such interval.

 

Summary: 

1.) CommVault asks vCenter to take a snapshot of the VM,

2.) Nimble array discovers the snapshot and puts in the queue for replication

3.) CommVault finishes and asks vCenter to delete the snapshot

4.) If the Nimble array has not finished replicating the snapshot, it refuses to delete it.

5.) vCenter on its side is also done with it and doesn’t feel ownership anymore.

6.) The Nimble array finishes the replication and leaves the snapshot alone since it was triggered outside of the array

7.) Since the state was online, the snapshot remains online.

Hi ThomasG,

not sure if understood you completely. Are you using IntelliSnap for Nimble snap engine with the VSA for VMware agent?

If so, the issue you see should not happen because in that case the process is as follows:

1.) CommVault asks vCenter to take a VMware snapshot of the VM

2.) As soon as the VMware snap creation is confirmed by vCenter, Commvault asks the Nimble array to create a hardware snap of the Datastore volume where the VMware snapshot is located on.

3.) Nimble array creates the hardware snapshot and puts te volume into the queue for replication

4.) CommVault finishes and asks vCenter to delete the VMware snapshot

5.) The hardware snapshot will be kept according to its retention (specified on the pimary snap copy of the Storage Policy)

6.) Once the retention for the hardware snap has passed, Commvault will ask the Nimble array to delete it

With this process there is no conflict of snaps since the hardware snap is still containing the VMware snap that will be replicated by the Nimble array, and it won’t be deleted until the retention has passed, even if the VMware snap was already deleted.

So, I don’t understand the issue you are seeing as that shouldn’t happen at all.

 

 

 

 

 

Likewise as to what @MarkusBaumhardt said - its unclear if you are using IntelliSnap or not. For nimble to pick up a snapshot, it sounds like its a hardware snap that gets created on the volume - but that could either be that you are using VVOLs or IntelliSnap.

If you are using IntelliSnap with the nimble snapshot engine, you may want to consider disabling the management of the replication from the Nimble side, and make Commvault aware of the replication on our side using these instructions:

https://documentation.commvault.com/2022e/expert/99655_configuring_hpe_nimble_storage_snapshot_replication_feature.html

Hello Markus,

I hope you are fine. Thank you for your answer. The question was from a customer and he doesn’t  want to spend more time on this, I don’t know why. Last response was:

I think its worthwhile, I can monitor the results on the storage side

Best regards,

Thomas

Reply


Terms & ConditionsCookie settings