Backup misisng on tape job?

[plandata_at]

Hi!

Using Veeam B&R9 for a Backup2Tape Job, Source of the tape job is a scale out backup repository, full backups shedules every Monday, job running at 8 AM every Monday, (this time job wa s manually started on wednesday)
So job closed with result success, no errors, no warning.
I decides to do a check if everything on tape and was shocked to see that instead of the success message complte fileserver is missing on tape! Log of the backup job says:

Code: Select all

29.06.2016 16:53:32 :: Building source backup files list started at 29.06.2016 16:53 
29.06.2016 16:53:32 :: No restore point for 27.06.2016 is available for virtual full backup, using previous restore point for 25.06.2016 
29.06.2016 16:53:32 :: Full backup from ZUCGRPSV151.vm-86382016-06-25T212553.vib (27.06.2016 00:00:00) restore point will be placed to media set 
29.06.2016 16:53:32 :: Source backup files detected. VBK: 0, VBK map: 1, VIB: 0 
29.06.2016 16:53:33 :: Creating full backup map for ZUCGRPSV151.vm-86382016-06-25T212553.vib 
29.06.2016 16:55:17 :: Queued for processing at 29.06.2016 16:55:17 
29.06.2016 16:55:19 :: Processing full backups started at 06/29/2016 15:47:50 
30.06.2016 07:56:15 :: New tape backup session started, encryption: disabled 
30.06.2016 07:56:18 :: [ZUCGRPSV151] Processing ZUCGRPSV151.vm-86382016-06-25T212553.vbk 
30.06.2016 08:09:45 :: 0 directories and 0 files backed up 
30.06.2016 08:09:45 :: Busy: Source 0% > Proxy 25% > Network 0% > Target 83% 
30.06.2016 08:09:45 :: Primary bottleneck: Target 
30.06.2016 08:09:45 :: Network traffic verification detected no corrupted blocks 
30.06.2016 08:09:45 :: Processing finished at 30.06.2016 08:09:45 

Whats wrong here and why is Veeam B&R not backing up this VM without even one warning?

[plandata_at]

Case ID: 01841654

[Shestakov]

Hi,
There are no errors or warnings because the file was written to a tape.
How often does the source job run? Is that a single or several backup jobs?
Could you check what is the most recent backup of the server on the scale-out repository?
Thanks

[plandata_at]

There is no errors or warnings because the file was written to a tape.

Im my opinion the file is not written to tape--> 0 directories and 0 files backed up

How often does the source job run? Is that a single or several backup jobs?

Source jobs are 18 backup jobs, running once per week on friday and finished on saturday night or sunday morning, the have all completed successfull.

Could you check what is the most recent backup of the server on the scale-out repository?

On Scale out repository there are three files of the VM:
- .vbm file from 25.06. 21:58
. vbk file from 25.06 21:58
. vib file from 25.06. 21:58

When looking at the restore points in GUI, is says Increment restore point on 25.06.2016 21:45 und full from 18.06.2016 21:46.

Thanks!

[Dima P.]

Hi plandata_at,

Can you check the properties (right click the tape > Properties > Files tab) of the tape media used during this job run?

[plandata_at]

Hi Dima,

looked on the propertied of all Tapes used during this backupjob, missing fileserver is not there.

[veremin]

Keep working with our support team - logs investigation should shed some light on the root cause of the experienced behaviour.

Our of curiosity - what product version are you currently at (aren't you using 9 GA by any chance; there used to several problems with virtual full candidate selection logic that have been fixed in Update 1). Wasn't the said job locked at the time when tape job was executed?

Thanks.

[plandata_at]

Using B&R 9.0.0.1491
Support Team has escalated case yesterday.
What do you mean by that the job was locked?

By the way: What design would be better: Do create virtual full by the tape job (one single job) or would it be better to create synthetic full by every source job, or just make incremental by source job and set restore pints to 1?. (source jobs are several jobs running about 3 days before the tape job). In the moment source jobs is incremental source job with 2 restore points. Since we are using storage snapshots for primary "backup", i only need one full per week on source jobs and on tape. Do you have any suggestions how to design this best?

[plandata_at]

By the way: What design would be better: Do create virtual full by the tape job (one single job) or would it be better to create synthetic full by every source job, or just make incremental by source job and set restore pints to 1?. (source jobs are several jobs running about 3 days before the tape job). In the moment source jobs is incremental source job with 2 restore points. Since we are using storage snapshots for primary "backup", i only need one full per week on source jobs and on tape. Do you have any suggestions how to design this best?

Just to bring up this topic again: Do you have any design recommondations for this? THX!

[plandata_at]

Hi!

Have the answer from Veeam support in the meantime. Unfortunatly this seems to be some sort of bug, but can be overcome by selecting jobs as a source instead of repository. I will test this next monday.

It turns out the issue have quite deep internal logic involved so it is quite difficult to fix it in the current release. However the issue is reported to our developers and Q&A. With this we will be able to make sure the issue is going to be fully tested prior the next major release.

Currently as a workaround I can suggest to add jobs as a source to the tape job, but not the whole SOBR repository. We could not reproduce the issue with this scenario.

[Shestakov]

Have the answer from Veeam support in the meantime. Unfortunatly this seems to be some sort of bug, but can be overcome by selecting jobs as a source instead of repository. I will test this next monday.

Thanks for updating!

By the way: What design would be better: Do create virtual full by the tape job (one single job) or would it be better to create synthetic full by every source job, or just make incremental by source job and set restore pints to 1?. (source jobs are several jobs running about 3 days before the tape job). In the moment source jobs is incremental source job with 2 restore points. Since we are using storage snapshots for primary "backup", i only need one full per week on source jobs and on tape. Do you have any suggestions how to design this best?

I would choose forever incremental backup method for primary job and virtual full backups for tapes, because it will use less space.

[plandata_at]

Hi!

Have the answer from Veeam support in the meantime. Unfortunatly this seems to be some sort of bug, but can be overcome by selecting jobs as a source instead of repository. I will test this next monday.

Selected jobs as a source now, and this workaround seems to solve the issue!

[plandata_at]

Do you have any informations is this is fixed in 9.5 now?

[Shestakov]

Yes, that is fixed in v9.5