Bug #25333 | ERROR 1032 (HY000): Can't find record in '<table>' | ||
---|---|---|---|
Submitted: | 29 Dec 2006 17:08 | Modified: | 31 Aug 2007 16:43 |
Reporter: | Matthew Boehm | Email Updates: | |
Status: | Can't repeat | Impact on me: | |
Category: | MySQL Cluster: Cluster (NDB) storage engine | Severity: | S1 (Critical) |
Version: | 5.0.27 | OS: | Linux (RHEL4) |
Assigned to: | Hartmut Holzgraefe | CPU Architecture: | Any |
Tags: | can't, cluster, error, join, record |
[29 Dec 2006 17:08]
Matthew Boehm
[29 Dec 2006 17:12]
Matthew Boehm
Changed 'custom' column to CHAR() makes no difference. Adding INDEX to 'custom' makes no difference.
[29 Dec 2006 17:41]
Matthew Boehm
Say what? This works? SELECT s.product_id, s.quantity, s.publicComments, s.privateComments, s.custom, p.product_name, p.product_parentid FROM subscriptions s LEFT JOIN products p USING(product_id) WHERE s.companyID = 1 ORDER BY custom, p.product_parentid ASC; Yep. Sure does. What the heck is going on? If I remove that 2nd ORDER BY column the query fails.
[2 Jan 2007 10:46]
Hartmut Holzgraefe
Hello Matthew, i was not able to reproduce this yet, are you able to provide table data showing the wrong behavior for reproduction on our side?
[2 Jan 2007 17:42]
Matthew Boehm
I'll provide anything you want. Tell me exactly what you need and I'll provide.
[2 Jan 2007 17:56]
Hartmut Holzgraefe
Hi Matthew, we'd need a mysqldump of your "products" and "subscriptions" tables. You can upload the dump files as private files only visible to MySQL staff using the "Files" tab in the bug form: http://bugs.mysql.com/bug.php?id=25333&files=1 or upload them to our FTP server at ftp://ftp.mysql.com/pub/mysql/upload/ if their size exceeds 200KB
[9 Jan 2007 10:57]
Cass Surek
Hello Hartmut, We are having the same thing with a very simple select with an ORDER BY from a table. Can I help in any way towards the solution for this matter? Cheers Cass
[15 Jan 2007 15:00]
Dan Bress
I am also seeing this issue. We are running mysql 5.0.24a. Our table is an InnoDB table where multiple processes are simultaneously doing update order by queries. The problem comes and goes, some connections get the error while others don't.
[3 Feb 2007 0:00]
Bugs System
No feedback was provided for this bug for over a month, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open".
[3 Feb 2007 18:53]
Matthew Boehm
Just reopening the bug per the automated system. I gave everything asked for. Waiting on developer response.
[7 Feb 2007 21:11]
Cass Surek
Since it takes this long for the problem to be dealt with, we are dropping the clustering and using drbd. Maybe our next attempt will turn out better.
[8 May 2007 8:52]
Jonas Oreland
Hi Matthew I retested your sql-statements with the data you provieded. I tested on 5.0.38 I tested with a 1-node cluster. They all worked fine. /Jonas
[31 Aug 2007 16:43]
Hartmut Holzgraefe
Same here, tested with both 5.0.27 and a current 5.0 version, with both engine_condition_pushdown enabled and disabled, on a 1, 2 and 4 data nodes configurations but was never able to reproduce the problems with the provided dumps and queries ...