Hello everyone,
I have a large table (around 15m rows) I'm trying to Full-Text index using
the Japanese wordbreaker.
When I do a full process, after processing 3-4 million rows, I see a couple
of "Error Fetching URL (80070006 - Invalid handle)" records in the Gather
Log, followed by many many c0000005 records in the log.
At that point, the MSSearch service is still running, but if I try to shut
it down using the sql server service manager or the windows tasks control
panel, it hangs while stopping. I have to kill it using the task manager
process panel.
I tried applying SP4 and recreating the catalog, but I have the same problem.
Any suggestions?
Regards,
Jonathan
Hilary Cotter - 19 May 2005 15:00 GMT
does this apply -
http://support.microsoft.com/default.aspx?scid=kb;en-us;298794

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
> Hello everyone,
>
[quoted text clipped - 17 lines]
>
> Jonathan
Jonathan Levine - 20 May 2005 06:53 GMT
Hilary,
Thanks very much for your reply.
> does this apply -
> http://support.microsoft.com/default.aspx?scid=kb;en-us;298794
I don't think so. I have increased MaxPropStoreCacheSize gradually to 0x0460.
Regards,
Jonathan
Hilary Cotter - 20 May 2005 17:25 GMT
This is possibly a buffer overrun. How are you storing your data? Is it in
char, varchar, text or image? If Image what is your document format?

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
> Hello everyone,
>
[quoted text clipped - 17 lines]
>
> Jonathan
Jonathan Levine - 20 May 2005 23:48 GMT
Hi again Hilary,
> This is possibly a buffer overrun. How are you storing your data? Is it in
> char, varchar, text or image? If Image what is your document format?
Buffer overrun sounds likely to me.
The data is in two columns: one big nvarchar, and an ntext.
It is Japanese text from web pages (the web pages have been stripped of HTML
tags).
Regards,
Jonathan
Hilary Cotter - 21 May 2005 01:37 GMT
just out of curiosity, do you get the problem when you index the content
with the neutral word breaker?

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
> Hi again Hilary,
>
[quoted text clipped - 13 lines]
>
> Jonathan
Jonathan Levine - 21 May 2005 05:57 GMT
Hilary,
> just out of curiosity, do you get the problem when you index the content
> with the neutral word breaker?
I haven't tried the neutral word breaker. The neutral word breaker doesn't
handle Japanese text very well: most Japanese is written without whitespace
between the words, so I don't think the neutral word breaker index would be
very useful even if it worked.
Jonathan
Hilary Cotter - 21 May 2005 10:29 GMT
I realize you are loosing the language capabilities you need - however if it
works using the neutral word breaker the problem is with the Japanese word
breaker and you can raise this as a bug with MS. If it fails with the
neutral word breaker something is wrong with your environment which might be
solved with a reapplication of the service pack, or by removing and
reinstalling SQL FTS.

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]
>
> Jonathan
Jonathan Levine - 27 May 2005 14:14 GMT
> I realize you are loosing the language capabilities you need - however if it
> works using the neutral word breaker the problem is with the Japanese word
> breaker and you can raise this as a bug with MS. If it fails with the
> neutral word breaker something is wrong with your environment which might be
> solved with a reapplication of the service pack, or by removing and
> reinstalling SQL FTS.
Hilary,
Your instincts were right on. The problem was in the Windows 2003
wordbreaker -- upgrading to SP1 fixed the problem. (see KB article 885899)
Thanks for your help!
Jonathan
http://kerblog.com/earlyedition/archive/2005/05/27/460.aspx