Bug #37530 Seemingly random data corruption on tables with triggers
Submitted: 19 Jun 2008 19:07 Modified: 20 Jul 2008 19:50
Reporter: Chris Vandomelen Email Updates:
Status: Closed Impact on me:
None 
Category:MySQL Server: Stored Routines Severity:S2 (Serious)
Version:5.1.23-rc-community OS:Linux (RHEL 5.1)
Assigned to: CPU Architecture:Any

[19 Jun 2008 19:07] Chris Vandomelen
Description:
I have a series of tables with triggers on those tables to preserve the former contents of the row, such that we don't lose data in those tables by the rows being changed (we have the full previous state of each row, in theory).

One of these tables, for no apparent reason, seems to be getting slightly corrupted on certain field updates; specifically, 2 fields seem to be picking up random segments of various other records, occasionally including itself, and overwriting important information within the record.

The server is a RHEL 5.1 server, running a kernel with version "2.6.18-53.1.14.el5PAE" on a quad-core Xeon X3210 (32-bit mode); the client software is Apache 2.2.3 in prefork mode, PHP 5.2.5, and an e-commerce system written in-house.

I would suggest a mechanism for repeating this bug, but it only seems to happen at random; the only reason I feel fairly certain that it's not a bug in the client code is because I logged the queries generated by the client code, and those fields were not changed.

How to repeat:
Unknown; happens randomly; looks like a buffer overflow or lack of initialization?
[19 Jun 2008 19:09] Chris Vandomelen
Data structure, data, queries

Attachment: bug detail.txt (text/plain), 14.82 KiB.

[19 Jun 2008 19:10] Valeriy Kravchuk
Thank you for a problem report. Please, try to repeat with a newer version, 5.1.25, and inform about the results.
[19 Jul 2008 23:00] 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".
[20 Jul 2008 19:50] Chris Vandomelen
The issue doesn't seem to be occurring any longer after making an update to 5.1.25.