BE9.1 for Windows constantly fails with unexplained errors

[andycole]

Hi all, this is my first post here.  Hopefully my last, if you get my meaning!

We have BackupExec 9.1 for windows on a server which backs up 5 servers including itself, to an Ultrium drive.  In the last couple of weeks, we've had a few failures, and they gradually got more frequent, and now it fails every day.  I looked up the error codes and found ZERO about them so I'm kind of stuck. 

The backup fails at a different point each day, so it's not a particular drive or server that is causing the problem. 

The initial error was:0x6a88 (27272) other error, which happened a few times, every few days,

and then last Thursday and Friday we got:ofo:Snapshot error:(0xfffffedl) A failure occured querying the writer status. I wass able to find out some info about that one which suggested rebooting the target server, which I did.

Since then, we have got: 0x5f88 (24568) other error.

Does anyone have any ideas about what this might mean?

Thanks

Andy

[ZeRoC00L]

Please not that BE9.1 is already very old, and as far as I know not supported anymore by Symantec.

Therefore Symantec does not test this version anymore to new service packs and hotfixes that are released by Microsoft. Also this version cannot support the new server OSes like Windows 2008.

It can be a wise idea to upgrade to a recent version of BE, like 2010R3.

[andycole]

Thanks for the response.  I did expect someone to say something like that, and I understand the response, but I don't think that compatibility with hotfixes is the cause of these errors, because we have had some successful backups in between the failures.

Andy.

[ZeRoC00L]

Have you also seen these technotes?

http://www.symantec.com/business/support/index?page=content&id=TECH27875

http://www.symantec.com/business/support/index?page=content&id=TECH42921

[CraigV]

Hi Andy,

Not sure what OS your servers are running as you never mentioned this.

however, snapshot erros can point to an issue with VSS if Windows Server 2003/2008. I'd suggest running: vssadmin list writers on your media server, and the server/s giving you that issue. All VSS writers should be stable and running. If they're not, then they're faulty. Normally a reboot fixes this, but if you're running Windows Server 2003, then consider repairing your VSS *.dlls using the link below:

http://support.microsoft.com/kb/940184

There is a small utility you can download and run which reregisters your VSS *.dlls.

I'd also check out your backup device...make sure you're not getting any hard errors under the Devices tab, and that no hardware errors are being logged in Event Viewer.

Thanks!

[andycole]

Thanks for the reply, CraigV. It's been some help.

We have a variety of versions of windows, a mixture of 2000 and 2003. I was aware of the snapshot problem and a reboot of the offending server did fix that one, but the other errors still remain.

I looked at the event log and at the same time as the last error, guess what? Backupexec was unable to update the catalog....due to insufficient disk space!    Now, the data on the drive where the catalogs are is usually pretty static, in fact, the backupexec data is about all that's on there that can grow. There are about 2GB of catalog files on there going back several years.  Whats the proper way to get rid of these?

I'm not sure if that is the whole solution, as the previous errors didn't coincide with a full HD, but we'll wait and see once I've cleared some space.

Andy

[andycole]

I've read those articles, thanks.

Like I said, I rebooted the server which made the vsswriters stable once again. They are still stable, so that is not what caused the most recent error.  It looks like I have had 3 different problems:

24568 caused by insufficient disk space

The snapshot error caused by a failed snapshot writer, which is now stable

The original 27272 errors, which I have no solution for, yet. There are no hardware faults showing or events in the event log, apart from the backup failed alert.

Andy

[CraigV]

...you can look at deleting your old catalog files, but if you ever need to restore from a tape whose catalog was deleted, you need to recatalog that tape first...

[andycole]

Hi again

I have deleted some old catalog files because I had no choice. I did it in windows explorer.  But surely there's an official way to do it from within BE?  I tried setting the 'truncate catalogs' option to truncate catalogs older than 6 months, this is supposed to happen during database maintenance at 5:00am, but nothing has been deleted.

On the backup issue, it looks like the problem is all to do with the VSS service on another server.  I ran a backup excluding the system state on that server and it went OK.  So now I've got to look for a solution to that problem.  It would appear that all the hotfixes for it are for W2003 SP1, but I have SP2, so surely that would include those hotfixes, wouldn't it?

Andy

[CraigV]

...selecting to truncate logs at 6 months will start that from the moment you set it. It won't truncate existing logs.

You deleted the logs correctly.

[andycole]

Thanks for that, at least I can fix that particular problem.

About the VSS writer problem, like I said, all the solutions I've found appear to be aimed at pre-SP2 versions of W2K3. We have SP2 so I would think that would include those patches.

Yesterday, during the day, I did a tape backup of everything we normally do, but excluding the exchange database and system state on the affected server.  That went through OK. So, as an experiment, I changed the selections for the overnight backup to do exactly the same thing, and that failed.  Well, it didn't even start. The job monitor showed a status of 'running' but there were 0 bytes backed up. I canceled the job, which didn't actually cancel until I'd killed the job engine service! When I went back in, the job log showed nothing other than that network communication had been established with the server.  My best guess is that even though I wasn't backing up the system state, the failed VSS writers were somehow still stopping the backup from proceeding.  At the previous backup I took during the day, the writers would have been stable due to a previous reboot.

This is confusing the hell out of me. Is there a different fix for the VSS issue if you have W2K3 SP2?

Thanks

Andy

[CraigV]

...the patch in the link I posted reregisters the VSS *.dlls. I've run it on a couple of my servers now, and it seems to work a treat.

The least you can do is download and run it. You can restart if need be, but so far I haven't had too. Do so, and see if you get the same issue on the server in question!