Highlighted

Vmware grt restore is staging forever

I have some issues when trying to restore a file from a vmware GRT backup. Backup is working without giving any errors, and when i browse the backupsets i can see all files including the ones i try to restore. I select the files i want to restore and just click "run now" (dont know if i'm supposed to make some special settings, but i cant really find anything that looks wrong).

Anyway, the restore starts with staging (we take our backups to a Virtual Tape Library so i know that staging is normal). I can follow the progress i the staging folder. I get a folder called IMG000001 and inside that one it is restoring the vmdk file.

So far everything is fine, the problem comes when the vmdk-file is done. Then i get another folder called IMG000002, and in that one it creates a new vmdk file. Then it makes another folder called IMG000003 and it continues until my stagingdisk is full and the restore fails.

I havent been able to figure out why it is doing like this. Has anyone else seen this? Do i have some settings wrong?

 

Also, if i look at the restore job, and go to the "job history" tab, i get a very long list that looks like this

Set Detail Information - Restore \\my-file-server.my.domain.com\g:

Set Detail Information - Unknown \\my-file-server.my.domain.com\g:

Set Detail Information - Unknown \\my-file-server.my.domain.com\g:

Set Detail Information - Unknown \\my-file-server.my.domain.com\g:

.......

It's about a 100 lines long i think. Cant really figure out what it means.

regards

 

/Henric

 

8 Replies

Did you do Incremental

Did you do Incremental backups?

Does the server you are trying to recover contain an Exchange or MSSQL Backup Set with GRT enabled?

 

Potentially you might need more disk space then you think for the staging area for either above scenario.

 

 

I do incremental backups, but

I do incremental backups, but i think i've had the same problem with restoring from full backups ( i take full once every two weeks and incremental daily). I've also had the same problem when only restoring one file, and i see no reason for it then, to use more than one backupset.

The server is only a file server with no other applications on it.

 

Ok so to restore one file

Ok so to restore one file from a Vmware incremental you have to stage - the full set, and every incremental set - so 13 days worth of incrementsla plus the full is worse case scenario for the staging requirements to get 1 file back with you (apparent) configuration.

Reason we need all of them back is that an AVVI backup is a block level backup so how would we handle the scenario where a large file that changes daily is involved with blocks from within the file located in the full and every incremental since the full.

Note 1: A file level incremental backup would only need the incremental - but this is block level and things work differently for block level operations.

Note 2: The full backup will still stage but (unelss Exchange or SQL GRT is involved) should only create the one IMG folder.

 

Would recommend

1) Change to B2D backups (will need more local disk space) instead of VTL - duplicate to VTL/tape if needing a secondary copy. This will eliminate the need to stage so make the restore quicker.

2) Provide more disk space to the staging area and let the jobs actually complete

3) Increase the frequecny of the full (say weekly) to reduce how many incrmental sets are needed to be staged.

 

 

The restore now failed due to

The restore now failed due to out of diskspace. It created 59 folders in my staging-area  (last one is IMG000059) which i really cant understand.

If i try to restore a file from a full backup it should only need to stage that full backup-set i guess? Then my staging-area should be more than enough. I will try that and report back my results.

 

 

Did you have one or more full

Did you have one or more full backup fail meaning the incremental backups are chained for more stpes than you would expect?

I have done some more tests

I have done some more tests and have to ask again about how it does on restore.

I tried a restore from a full backup. My vmdk disk is about 550 GB unfortunately so it takes a while, but that one worked fine. It restored the vmdk file and got the file out of it and all was fine. One thing though. After taking the full backup, i duplicate it to physical tape. When restoring from the full backup, it restored from the physical tape even though the Virtual Tape was chosen as device in the restore job which i really cant figure out why.

 

I then tried to restore a file from the backup the day after the full backup. From my understanding it should then need the full backup and the incremental backup, which should fit on my staging drive which have 1.5 TB free space. Today when i got to the office it had failed with error "not enough space on staging device".

Should it be enough with those two backups or does it need more space? Does an incremental backup still take the space of a full when restoring it?

 

When trying to understand the job-log it looks a bit strange. I get 6 blocks of "Set information". Is each one of those a restore of a backup-set?

The first 4 blocks is about my full backup. From what i can see it looks like it first restores the duplicate clone, then the original backup, and then tries to restore the clone again, and then another one of the original. Which make a total of 4 copies of the full backup. My diskspace then runs out, but after that it was going to restore the incremental set. How come it restores so many copies of the full backup?

Also, when looking at my earlier restore which failed, I have about 50 "Set information"-blocks about my clone, and then 50 more about my full backup. If those are needed i will never be able to restore from my backups.

I have changed the backup of the fileserver so i'm now using normal windows agent and taking backup over LAN, but i still have other servers set up for vmware backup and just want to be prepared if i ever need to restore from those backups.

 

Just a thought. Are you using

Just a thought. Are you using the Change Block Tracking on your VM's? if you are not then you are taking a full backup everyday even though you have the Differential/Incremental setting. VM's must have CBT enabled for the Incremental/Differential option to work. That may account for your running out of drive space. Trying to stack multiple full VM backups on top of each other. You will run out of room very fast.

Suppose you have backup sets,

Suppose you have backup sets, Full, Incr1, Incr2 and Incr3 and you want to restore to Incr3.  Did you select all the backup sets, i.e., Full, Incr1, Incr2 and Incr3, or did you just select Incr3?  You are supposed to select only Incr3 and BE will pull in the other backup sets that it is dependent on.