Problems with x64 backups

[Rob_Baumstark]

Ok - we'll start with the current configuration. The media server is Win2K3 SP1, all up to date on security fixes, BackupExec 10d with hotfix 1 applied. The servers I'm trying to back up are Win2K3 Enterprise x64 edition in a MSCS cluster with SQL Server 2000 SP4 and the x64 AWE hotfix applied. There are 4 virtual SQL instances running on 2 physical nodes.

And the problems...

First off, I can't see any of the SQL instances from BackupExec. Yes, I have set RestrictAnonymous to 1. The documentation on whether SQL 2000 is supported on Windows x64 editions is contradictory, so I'm not even sure what to think about this. As a workaround I currently have the SQL servers doing backups to local disks, and BackupExec backing up the SQL backup files (or trying to).

Next problem - the backup jobs take much longer than they should to start - about 20 minutes. During this time nothing is happening except errors being generated in the windows event log of the server being backed up. After about 20 minutes, the job will finally start processing data, it actually runs at a decent speed, about 1GB / minute. There are no other jobs running, and the backup destination is a B2D file so it's not waiting for a tape to seek to the proper position or anything. The errors in the event log of the server being backed up are VSS error #6013, related to the VSS provider for SQL. The exact error is the same as the one at this MS article right near the bottom: http://support.microsoft.com/default.aspx?kbid=833167 however unlike the text of that article, we have SQL 2000 SP4 Enterprise instead of MSDE, and we are doing a backup of the NTFS filesystem, not a restore of a database. Because I can't even see the SQL databases in backupexec (and have not selected them) there is NO reason for the remote agent to be initializing the SQL VSS provider. And because I am not licensed for the AOFO, I can't select a different snapshot provider to see if it would help.

Last problem - since installing hotfix 1 for 10d and adding a few Windows x64 based servers to some of my backup jobs, bengine.exe has been crashing every night. I'm currently working on narrowing down the scope of this problem, because I can run backups of the x64 servers successfully, and it doesn't necisarrily crash during one of the x64 jobs. However it didn't start until after applying hotfix 1, which according to the release notes does not modify any files on the media server except to store the installer for the x64 agent. This crashing is causing major problems in our production backup thuogh, as any tapes in use at the time have corrupted end-markers, and I'm having to re-run a lot of jobs, using up a lot of extra disk and tape space.

[Rob_Baumstark]

Just crashed the backup engine again - this time it happened during a backup duplication job. The message from the windows event log is:

Faulting application bengine.exe, version 10.1.5629.0, faulting module bengine.exe, version 10.1.5629.0, fault address 0x000e9ad5.

I've also configured the service to restart itself if it crashes, which BackupExec seems to have problems understanding as well. BackupExec is claiming that the user 'Recovery' cancelled the job that was running during the crash - there is however no such user.

[tejashree_Bhate]

Hello,



Please launch the sgmon and then start the services and paste the contents of the sgmon as it would be easier for troubleshooting this issue
Also please check the application and system event logs for errors

Note:- if we do not receive your reply within two business days, this post would be assumed answered and archived.

[Rob_Baumstark]

Another crash last night. Same error as last time in the app event log on the media server:

Faulting application bengine.exe, version 10.1.5629.0, faulting module bengine.exe, version 10.1.5629.0, fault address 0x000dbd05.

Just before it happened, this error was also entered in the event log.

Storage device "Local Array" reported an error on a request to write data to media.

Error reported:
The physical end of the tape has been reached.
.

The "Local Array" device is a B2D folder. It is configured to always leave 10GB free on the disk, so it's not a disk-full problem, however it has used all of the space that it is allowed to use. Running out of space shouldn't cause a media error or crash the backup engine though - the job should simply continue on spanning media to the other B2D folders in the pool.


The only error messages I'm seeing in the event logs of the servers I'm backing up, is on the x64 servers the logs are full of this message:

Sqllib error: OLEDB Error encountered calling IDBInitialize::Initialize. hr = 0x80004005. SQLSTATE: 08001, Native Error: 17
Error state: 1, Severity: 16
Source: Microsoft OLE DB Provider for SQL Server
Error message: SQL Server does not exist or access denied.

Which I think I've tracked down to be a bug in SQL 2000 when installed on Windows x64. Waiting for the MSDE VSS provider to time-out (which results in that error) is what is causing backup jobs from x64 servers to take almost 20 minutes before they start doing any work. And though I do believe it is a problem that Microsoft will have to solve, there is still no reason for the backup agent to be touching the MSDE VSS provider, or any VSS provider. I'm doing filesystem backups only (no system state), and no advanced open file option, from the x64 servers.


I'll play with sgmon in the next couple days sometime, I've also got other issues to work on here.

[shweta_rege]

Hello,

- Please launch the sgmon and then start the services and paste the contents of the sgmon as it would be easier for troubleshooting this issue


******************************************************************
*****************************************************************

Note : If we do not receive your reply within two business days, this post would be marked �assumed answered� and would be moved to �answered questions� pool.


Thanks.