Bad Byte Images
A site running IDS 7.31UD4 on Solaris had a huge table with a BYTE column in its own blobspace; oncheck reported 17 bad blob images and the owners asked whether deleting those rows would be safe or might crash the server. Suggestions were to copy/test on another host first (impractical here, 226 2GB chunks), and that deleting by rowid should work, or trying to simply update the blob. oncheck -cDI produced occasional assertion failures (af files) but the server stayed up. Art Kagel explained each row holds a 56-byte blob locator in the tablespace pointing to the first blob page (see locator.h). No confirmed outcome of the delete is recorded.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
Gurus, We have an extremely large table (millions of rows) that has one of its columns defined as a byte column in a separate dbspace. Oncheck indicates that there are 17 bad images. We know which rows contain the bad images. Would it be safe to just delete those rows or do we risk "crashin" the system? IDS 7.31UD4 on Solaris 5.8. Thanks in advance, Sam
If I were you, I would make a copy and test it on another host first. Frank SAMUEL JAKABOWSKI wrote: >Gurus, > >We have an extremely large table (millions of rows) that has one of its >columns defined as a byte column in a separate dbspace. Oncheck indicates that >there are 17 bad images. > >We know which rows contain the bad images. > >Would it be safe to just delete those rows or do we risk "crashin" the system? > >IDS 7.31UD4 on Solaris 5.8. > >Thanks in advance, >Sam > > >******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > > > -- Yunyao "Frank" Qu Computer Sciences Corporation(CSC) NOAA/CLASS, (301)817-4696
I would love to except there are 226 2GB chunks that make up this space. This is 7.31UD4 on Solaris 5.8. From: "Yunyao (Frank) Qu" <Yunyao.Qu@noaa.gov> Reply-To: ids@iiug.org To: ids@iiug.org Subject: Re: Bad Byte Images [7359] Date: Fri, 25 Aug 2006 17:07:02 -0400 (EDT) If I were you, I would make a copy and test it on another host first. Frank SAMUEL JAKABOWSKI wrote: >Gurus, > >We have an extremely large table (millions of rows) that has one of its >columns defined as a byte column in a separate dbspace. Oncheck indicates that >there are 17 bad images. > >We know which rows contain the bad images. > >Would it be safe to just delete those rows or do we risk "crashin" the system? > >IDS 7.31UD4 on Solaris 5.8. > >Thanks in advance, >Sam > > >******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > > > -- Yunyao "Frank" Qu Computer Sciences Corporation(CSC) NOAA/CLASS, (301)817-4696 ******************************************************************************* Forum Note: Use "Reply" to post a response in the discussion forum.
You 'should' be able to delete via rowid - it used to work back in the days of V4. Have you tried just updating the blob Paul Watson Tel: +44 1414161772 Mob: +44 7818003457 Web: www.oninit.com GO FURTHER with DB2 GET THERE FASTER with Informix. Attend the IDUG 2006 European Conference. Vienna, Austria. 2-6 October 2006 Visit http://www.iiug.org/conf for more information. -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Bob Allan Sent: 25 August 2006 16:43 To: ids@iiug.org Subject: Re: Bad Byte Images [7360] I would love to except there are 226 2GB chunks that make up this space. This is 7.31UD4 on Solaris 5.8. From: "Yunyao (Frank) Qu" <Yunyao.Qu@noaa.gov> Reply-To: ids@iiug.org To: ids@iiug.org Subject: Re: Bad Byte Images [7359] Date: Fri, 25 Aug 2006 17:07:02 -0400 (EDT) If I were you, I would make a copy and test it on another host first. Frank SAMUEL JAKABOWSKI wrote: >Gurus, > >We have an extremely large table (millions of rows) that has one of its >columns defined as a byte column in a separate dbspace. Oncheck >indicates that >there are 17 bad images. > >We know which rows contain the bad images. > >Would it be safe to just delete those rows or do we risk "crashin" the system? > >IDS 7.31UD4 on Solaris 5.8. > >Thanks in advance, >Sam > > >*********************************************************************** >******** Forum Note: Use "Reply" to post a response in the discussion >forum. > > > > -- Yunyao "Frank" Qu Computer Sciences Corporation(CSC) NOAA/CLASS, (301)817-4696 **************************************************************************** *** Forum Note: Use "Reply" to post a response in the discussion forum. **************************************************************************** *** Forum Note: Use "Reply" to post a response in the discussion forum.
We tried oncheck -cDI on the table. The error comes back but only
occassionaly. We get the af.X message but the system stays up. Data and images
on either side of the 17 are okay.
What we are unsure of is where is the pointer stored that links the byte back
to the table?
Every row that contains a BYTE or TEXT blob column contains a 56byte blob
'locator' as part of the actual row in table space. That is where the pointer
to the blob's first blob page exists along with other header information. Look
at the loc_t structure in $INFORMIXDIR/incl/esql/locator.h for locator column
contents (need the SDK installed to see this header file).
Art S. Kagel
----- Original Message -----
From: Samuel Jakabowski <ids@iiug.org>
At: 8/25 20:06:43
We tried oncheck -cDI on the table. The error comes back but only
occassionaly. We get the af.X message but the system stays up. Data and images
on either side of the 17 are okay.
What we are unsure of is where is the pointer stored that links the byte back
to the table?
*******************************************************************************
Forum Note: Use "Reply" to post a response in the discussion forum.