commvault.com commvault.com
Question

VSA restore issue from backup copy device

  • 19 October 2023
  • 3 replies
  • 43 views
J
Rank
Badge +5

Dear team

We need a confirmation or explanation for the following restore issue.

 

IntelliSnap Backups are running every 6 hrs.

Only the last daily “full” is selected to be backup copied to a disk library.

For NAS and Virtualization IntelliSnap backups, an incremental backup Job is interpreted as full.

As  the result the last IntelliSnap Backup of the day is selected to be backup copied.

 

Consultant did a Full VM restore from the copy available on the disk library but after the completed restore, the virtual machine did not come up.

 

Since only every 4th backup is moved to disk library, he’d expect, that an incremental backup copy uses the last completed backup copy as reference point for the backup copy.

But for him it looked as if the incremental backup copy is referencing the latest Snap Backup instead.

 

Please confirm:

  • Commvault needs the data of ALL incremental snap backups to enable a consistent restore.
  • An incremental backup copy can only be used for restores, if all snap backups, between the last full and the chosen point in time, have been transferred to the restore source media.
  • A incremental backup copy is always referencing to the previous backup job and not the previous backup copy job.

 

If this is correct, we have greater effort to change the configuration. Therefore, the question to chat.

Thanks in advance

Joerg

3 replies

J
Rank
Badge +5

Hi

It seems the explanation and question are not appealing enough.

But the topic is not uninteresting. Simpler:

VSA backup. First is a hardware snapshot. I do a full and a 10 incremental chain. In commcell console I see what I describe. 11 snaps, 1 full, Rest incr.

Now it is possible to configure the “backup copy” that it will backup every fourth snap to disk. That Means in my example I see one full and 2 incremental “backup copies” in Commcell console.

Now I do a restore. Complete VM. Copy precedence “backup copy”, snaps are no longer available. Recovery time is the second incremental. Recover does not fail but the vm is not useable.

Simplest explanation: I assume I need the 6 incrementals in between the gaps( each fourth snap to disk) to do a consistent recover. In the backup copy I have a broken chain.

Is that so?

My expectation: If I do a backup copy from hardware snap Commvault SW always take care that the backup copy is consistent.  In this case it should be a chain of one full and two incrementals. This should be possible because a snap is in reality always a full.

Can anyone explain/confirm the behavior?

I see problems because

  • the restore I made did not fail from Commvault perspective.
  • it is possible to configure.
  • in the next step we do a synthetic full, not knowing what happens

Thanks in advance

Joerg

L
Rank
Userlevel 5
Badge +13

Your second explanation it is easier to understand.

 

Backup copy will leverage the CBT information from the snapshot so I think your backup copy process is consistent and there is no broken chain.

 

My comments:

can you try to restore from the snapshot (or perform live mount) instead of primary copy for that specific VM to compare?

I think the fact that your VM does not boot once restored can be because the CBT status/info is not reliable. Reseting the CBT status and running new backups of the VM to test the restore again can lead to a differente outcome.

Additionally, this happen only to this VM or do you have any other VM to test the restore and validate they are usable.

 

 

J
Rank
Userlevel 1
Badge +7

Hi @Ledoesp,

we tried to restore from snapshot without any problems.

following Scenario:
1 - Full Snap + backupCopy 
2 - Incr Snap
3 - Incr Snap
4 - Incr Snap + backupCopy
5 - Incr Snap
…

when restoring from any Snap everything works fine.
restoring from backupCopy of backup #1 also works fine

restoring from backupCopy of backup #4 leaves an inconsistent VM (unable to boot)

same if Snap backups # 2 and #3 are aged. The restore from backupCopy #4 does not produce a propper VM.

worst about this is, that the restore process completes without any errors !
no missing information is reported.

In case you configure the backupCopy Policy to backupCopy only the first (or last) daily Full, and you create multiple Incremental backups a day (and continue to create only incremental backups), it is resulting in the above mentioned backup chain.

For NAS and Virtualization Clients, an incremental snapshot backup is handled like a full.

for me it looks like the backupCopy process does not refer to the CBT information used with the latest backupCopy, but to the CBT used with the latest (Snapshot)backup.
 

is there another way to create multiple backups per day, but only transfer one of the to a consistent primary copy (to disk, to cloud, to tape) ?

And making use of synthetic full instead of backing up the whole environment using full backups on 150+ TB of data ? (which is still increasing rapidly for this environment)

rgds
Klaus

 

Reply


Terms & ConditionsCookie settings