Home | Contact Us | FAQ | Search & Site Map | Link to Us
Sign In | Join | Other 45 Sites in Network
Home
Discussion Groups
DB Engine
SQL ServerMSDESQL Server CE
Services
Analysis (Data Mining)Analysis (OLAP)DTSIntegration ServicesNotification ServicesReporting Services
Programming
CLRConnectivitySQLXML
Other Technologies
ClusteringEnglish QueryFull-Text SearchReplicationService Broker
General
Data WarehousingPerformanceSecuritySetupSQL Server ToolsOther SQL Server Topics
DirectoryUser Groups
Related Topics
MS AccessOther DB ProductsMS Server Products.NET DevelopmentVB DevelopmentJava DevelopmentMore Topics ...

SQL Server Forum / Other Technologies / Full-Text Search / November 2006

Tip: Looking for answers? Try searching our database.

Surely a bug in SQL Server 2005 Full Text Search (FTS) or Backup?

Thread view: 
Enable EMail Alerts  Start New Thread
Thread rating: 
Richard Yeo - 13 Nov 2006 11:52 GMT
We are live (production system) running
- Windows 2003 Data Centre Edition R2 x64
- SQL Server 2005 Enterprise Edition x64 SP 1

The system has been operating fine for 7 days and all of a sudden we have
started getting this message every time we backup our database. The backup
fails!!!

We get this event in the logs...

Event Type:    Error
Event Source:    MSSQLSERVER
Event Category:    (2)
Event ID:    3633
Date:        13/11/2006
Time:        00:00:16
User:        <removed>\<removed>
Computer:    <removed>
Description:
The operating system returned the error '32(error not found)' while
attempting 'CreateFile' on
'F:\MSSQL.1\MSSQL\FTDATA\fish4_Z_Search_Freetext\SQL.HDR' at
'fulltext.cpp'(529).

followed by this event...

Event Type:    Error
Event Source:    MSSQLSERVER
Event Category:    (6)
Event ID:    3041
Date:        13/11/2006
Time:        00:00:16
User:        <removed>\<removed>
Computer:    <removed>
Description:
BACKUP failed to complete the command BACKUP DATABASE fish4. Check the
backup application log for detailed messages.

The full text index is set to Automatically track changes

According to Books Online...

===> During a full database backup in SQL Server 2005, full-text data is
backed up together with other database data.

During the backup, the catalog is put into a read-only mode, so that "crawl"
activity (the process creating and maintaining a full-text index) is
suspended until the backup completes. <===

So why are we getting this error message? Sure this is a bug in FTS (Full
Text Search) / SQL Server Backup?

Any assitance would be much appreciated as this is causing the backups of
our production system to fail therefore putting our business at risk. Our
business depends on FTS so we can't just turn it off.

Feel free to email me at...

richard_s_yeo at hotmail dot com

Regards
Richard
Hilary Cotter - 13 Nov 2006 12:48 GMT
We have had a very similar problem in RTM. We opened a support incident with
Microsoft on this one (SRX060201600683) and were told that it would be fixed
in SP1. They asked us to debug it (they gave us instructions on how to set
up the debugger) but we were unable to get a debug dump. We have applied SP1
and it is still not fixed. However our catalogs do not seem to get corrupted
the way they did previously.

Here is our error message:

The operating system returned the error '32(The process cannot access the
file because it is being used by another process.)' while attempting
'CreateFile' on 'G:\SQLFTS\SonarData_9\SQL.HDR' at 'fulltext.cpp'(529).

Note is is the same line number.

Signature

Hilary Cotter

Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html

Looking for a FAQ on Indexing Services/SQL FTS
http://www.indexserverfaq.com

> We are live (production system) running
> - Windows 2003 Data Centre Edition R2 x64
[quoted text clipped - 59 lines]
> Regards
> Richard
Richard Yeo - 13 Nov 2006 13:59 GMT
Hilary

Is this support case still open?

SRX060201600683

Do you have any contacts within the MS FTS / SQL Server team that might be
able to help?

Regards
Richard
Hilary Cotter - 13 Nov 2006 15:19 GMT
It is probably archived. My contacts on the full-text team are not helpful.

Signature

Hilary Cotter

Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html

Looking for a FAQ on Indexing Services/SQL FTS
http://www.indexserverfaq.com

> Hilary
>
[quoted text clipped - 7 lines]
> Regards
> Richard
Richard Yeo - 14 Nov 2006 23:33 GMT
It has failed 3 nights in a row now with the same error

We are experimenting tonight with a work around which pauses indexing before
the backup job and restarts indexing after the backup job

-- Pause FTS indexing
exec sp_fulltext_service @action = 'pause_indexing', @value = 1

-- Do backup

-- UnPause FTS indexing
exec sp_fulltext_service @action = 'pause_indexing', @value = 0

Not ideal!

Rich
Hilary Cotter - 14 Nov 2006 23:41 GMT
We do a file group back up. We have partitioned so we know that we can
rebuild our indexes within a day we are willing to live with the downtime
while our indexes are rebuilt.

Signature

Hilary Cotter

Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html

Looking for a FAQ on Indexing Services/SQL FTS
http://www.indexserverfaq.com

> It has failed 3 nights in a row now with the same error
>
[quoted text clipped - 13 lines]
>
> Rich
Richard Yeo - 15 Nov 2006 09:13 GMT
After the change to pause indexing before the backup and unpause indexing
after the backup has completed the backup didn't fail last night.

The jury is out as to whether this has really worked around the issue.

The Backup commands should automatically pause the indexing for the period
of time to backup the full text catalogs not for the entire backup period.
Our backup takes 30 mins of which 2 mins are probably spent backing up the
full text catalogs.

Regards
Richard
Steve Schmidt (SQL Dev) - 15 Nov 2006 19:24 GMT
We are aware of a problem when log backup and data backup are running
concurrently.
Can you determine if your log backups are running at the same time as your
failed backup?

A few suggestions:
- check the event log for log backups finishing around the same time as the
failed data backup
- use trace flags 3004, 3605 to see if there is overlap between the backups

The fix for this problem has not yet been released, but is in the pipeline
for the SP2 release.

Please contact SQL Support.  I agree that this is almost certainly a bug.
However, I can not confirm that the overlapped log/data backups problem is
the same as your issue.

Thanks,
-Steve Schmidt
SQL Server Developement

> After the change to pause indexing before the backup and unpause indexing
> after the backup has completed the backup didn't fail last night.
[quoted text clipped - 8 lines]
> Regards
> Richard
Richard Yeo - 16 Nov 2006 09:56 GMT
Steve

Thanks for taking the time to reply to my newsgroup post.

With fulltext indexing paused during backup the backup was successful for
the second night in a row. If it makes it to the end of the week I will
consider the workaround to have worked.

I can confirm that our backup is kicked off at 00:00 and that our
transaction log backups take place every 15 minutes starting at 00:00 so they
are running at the same time.

Is there a KB article number / bug ref i should quote when i contact
support. If there isn't a fix available i guess there is little point me
contacting support.

Are you saying that this has nothing to do with FullText?
Is our work around of pausing fulltext indexing the right thing to be doing?

Regards
Richard

richard_s_yeo at hotmail dot com
Steve Schmidt (SQL Dev) - 16 Nov 2006 20:34 GMT
I suspect your workaround is simply altering the timing, such that the log
backup is completed prior to the data backup starting.

FYI: The support case number is as Hilary pointed out:
SRX060201600683
Internally the bug number is 433064.
The fix is already in our product, but the fix has not yet been released.

For defects that are impacting your business, you can request a "QFE" which
can be released prior to the next service pack.

Another workaround would be to run with traceflag 3227.
That traceflag disables concurrent log/data backups.  Backups against a
given database serialize using "U" locks.

Thanks for your patience.
-Steve

> Steve
>
[quoted text clipped - 19 lines]
>
> richard_s_yeo at hotmail dot com
Richard Yeo - 17 Nov 2006 09:42 GMT
Steve

Thanks for replying again.

I like the sound of using a traceflag as apposed to applying a QFE although
this is affecting our business so we definitely need to get it fixed either
way and can't wait for SP2 to be released.

Are there any side effects from using the traceflag 3227, e.g. locking?

Users access our system 24 hrs x 365 days so we cannot afford for them to be
interrupted.

The only documentation on traceflag numbers i can find is here

http://msdn2.microsoft.com/en-us/library/ms188396.aspx

Regards
Richar
Steve Schmidt (SQL Dev) - 20 Nov 2006 22:07 GMT
There will be no locking impact for normal access.

Only the backups will block each other (as full backup and log backup did in
sql2000).

> Steve
>
[quoted text clipped - 15 lines]
> Regards
> Richar
Richard Yeo - 21 Nov 2006 09:43 GMT
Thanks

I changed the transaction log backup so it didn't happen at the same time as
the full backup until I had you confirmation. I also re-enabled fulltext
indexing during backup.

This seems to have cured the problem, i.e. it wasn't to do with FTS after all.

Regards
Rich
 
Sign In
Join
My Latest Posts
My Monitored Threads
My Blog
My Photo Gallery
My Profile
My Homepage

Start New Thread
Enable EMail Alerts
Rate this Thread



©2009 Advenet LLC   Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.