<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
OK, all is not as wonderful as I had hoped - it turns out that mysql
is smarter than I am. All of the relational integrity between the
original biblio table (which I renamed to biblio_orig) and other
tables was maintained. So, for example, the referential links
between biblio and biblioitems are now between biblio_orig and
biblioitems.<br>
<br>
Any ideas on how I fix this cleanly, re-linking everything to my
rebuilt biblio table?<br>
<br>
Thanks in advance!<br>
<br>
Chris<br>
<br>
On 3/10/11 9:29 PM, Chris Hobbs wrote:
<blockquote cite="mid:4D79B334.7050908@nhusd.k12.ca.us" type="cite">
<meta http-equiv="content-type" content="text/html;
charset=ISO-8859-1">
Hi all,<br>
<br>
Just got done pulling off a (very) minor miracle, but thought I'd
document it in the hopes that someone else in the same boat finds
it useful. <br>
<br>
We've been having performance problems for over a week, and I
finally got around to checking the correct logs, and discovered
that we had a corrupt page in one of those lesser used koha
tables, something called biblio :-/ <br>
<br>
Of course, since I'd let it slide for over a week, bringing biblio
back from backup would have meant lots of re-work for any
cataloging done since the corruption began.<br>
<br>
Because of the corrupt page, whenever a request was made against
that particular page, the mysql database would restart, dropping
the connections and causing general chaos. And thanks to google's
robot, that corrupt page was getting hit pretty regularly as it
turns out.<br>
<br>
Some googling gave me an idea of how to retrieve what I could from
the table (think binary search), finding the edges of the corrupt
page. I ended up being able to recover all of the biblio records
from 1-30227, and from 30313-91859, leaving a gap of 85 missing
biblios.<br>
<br>
I could have made our library technician (hi Katie) rebuild those
records by hand, but biblioitems has all of the required
information stored in the marcxml field. Thanks to prior questions
being asked and answered on the list, I was able to put together
the following query to recreate the missing records. <br>
<br>
insert into biblio<br>
select biblionumber, '' framework, <br>
extractValue(marcxml,"record/datafield[@tag='100'][1]/subfield[@code='a']")
author,<br>
extractValue(marcxml,"record/datafield[@tag='245'][1]/subfield[@code='a']")
title,<br>
NULL, NULL, <br>
extractValue(marcxml,"record/datafield[@tag='440'][1]/subfield[@code='a']")
REGEXP '[a-zA-Z]' series, <br>
extractValue(marcxml,"record/datafield[@tag='440'][1]/subfield[@code='a']")
series_title,<br>
substring(extractValue(marcxml,"record/datafield[@tag='260'][1]/subfield[@code='c']"),2,4)
copyright,<br>
NULL,<br>
'2010-07-23',<br>
extractValue(marcxml,"record/datafield[@tag='520'][1]/subfield[@code='a']")
abstract<br>
from biblioitems where biblionumber >= 30228 and biblionumber
<= 30312;<br>
<br>
Hope somebody else finds this useful!<br>
<br>
<div class="moz-signature">-- <br>
<font color="blue">Chris Hobbs</font><br>
Director, Technology<br>
New Haven Unified School District</div>
--
This message has been checked by ESVA and is believed to be clean.
<pre wrap="">
<fieldset class="mimeAttachmentHeader"></fieldset>
_______________________________________________
Koha mailing list <a class="moz-txt-link-freetext" href="http://koha-community.org">http://koha-community.org</a>
<a class="moz-txt-link-abbreviated" href="mailto:Koha@lists.katipo.co.nz">Koha@lists.katipo.co.nz</a>
<a class="moz-txt-link-freetext" href="http://lists.katipo.co.nz/mailman/listinfo/koha">http://lists.katipo.co.nz/mailman/listinfo/koha</a>
</pre>
</blockquote>
<br>
<br>
<div class="moz-signature">-- <br>
<font color="blue">Chris Hobbs</font><br>
Director, Technology<br>
New Haven Unified School District</div>
--
This message has been checked by ESVA and is believed to be clean.
</body>
</html>