Unhandled exception in backup

DBA_Dave

This may be a regression bug. It was fixed back at 3.2.0.10, according to a post in the forum.

There are about 70 more databases on the server, they all backed up without error. They are being backed up to a local SAN drive.

10/19/2007 12:00:33 AM: Backing up Calcs (full database) to:
10/19/2007 12:00:33 AM: r:\backup\Fri\FULL_SNLDB2_Calcs_2007.Oct.19.000033.sqb

10/19/2007 12:00:33 AM: BACKUP DATABASE [Calcs] TO DISK = 'r:\backup\<DATETIME ddd>\<AUTO>.sqb' WITH NAME = '<AUTO>', DESCRIPTION = '<AUTO>', VERIFY, ERASEFILES_ATSTART = 10, MAILTO_ONERROR = 'sqladmin@snl.com', COMPRESSION = 3, THREADCOUNT = 6

10/19/2007 12:08:21 AM: Thread 0 error:
Process terminated unexpectedly. Error code: -2139684860
10/19/2007 12:08:21 AM: Thread 1 error:
Process terminated unexpectedly. Error code: -2139684860
Process terminated unexpectedly. Error code: -2139684860
10/19/2007 12:08:21 AM: Thread 2 error:
Process terminated unexpectedly. Error code: -2139684860
Process terminated unexpectedly. Error code: -2139684860
10/19/2007 12:08:21 AM: Thread 3 error: Backup error: Access violation at address 10009801 in module 'zlib1.dll'. Read of address 062D6000
10/19/2007 12:08:21 AM: Thread 4 error:
Process terminated unexpectedly. Error code: -2139684860
Process terminated unexpectedly. Error code: -2139684860
10/19/2007 12:08:21 AM: Thread 5 error:
Process terminated unexpectedly. Error code: -2139684860
Process terminated unexpectedly. Error code: -2139684860

10/19/2007 12:08:21 AM: Database size : 57.997 GB

SQL error 3013: SQL error 3013: BACKUP DATABASE is terminating abnormally.
SQL error 3271: SQL error 3271: A nonrecoverable I/O error occurred on file "SQLBACKUP_2BEFB24B-C398-4862-8E60-EDC02EDC9E31:" 995(The I/O operation has been aborted because of either a thread exit or an application request.).
SQL error 3202: SQL error 3202: Write on "SQLBACKUP_2BEFB24B-C398-4862-8E60-EDC02EDC9E3102" failed: 995(The I/O operation has been aborted because of either a thread exit or an application request.)

Daniel Handley

We have received a few of these issues and it looks like there is a problem in the zlib library.
Currently we can't find out what the problem is as this is a third party dll and we don't have complete access to it.

The workaround at the moment is to reduce the compression level to 2.

If we find a solution we will release it as a patch or a minor version, otherwise we will be rewriting this for version 6.

Regards
Dan

DBA_Dave

ZLIB is free open source code. Everyone has access to it:

http://www.zlib.net/

(The Copyright strings are in your DLL's)

Daniel Handley

Dave

You are correct in that this is an "open" source project.
However the issue is that we cannot consistently replicate the error, and thus track down the issue within the zlib library.

Once we can find a way to make this fail every time we can resolve the issue.
The alternative we have is to change the compression algorithm. This will involve allot of work and testing so is unlikely to happen before version 6.

Regards
Dan

Keith_Walton

I recently increased my compression level to 3 - so I will change it back to 2. This problem only occurred once. I have my job set to auto-retry and it worked the 2nd time. I'm running version 5.2.0.2825 and I'm using 7 threads for this backup. I notice the original poster is using 6 - maybe the number of threads used can increase the likelihood of this error occurring. The database is 241 GB and I'm running on a 64-bit server. I am running multiple backups simultaneously.

I hope this information can help you reproduce the bug.