Timo Sirainen wrote:
I'm beginning to think that this has something to do with creating index files and a race condition somewhere in it. I've done a couple of fixes that should change this error message to something else:
http://dovecot.org/list/dovecot-cvs/2007-January/007375.html http://dovecot.org/list/dovecot-cvs/2007-January/007377.html http://dovecot.org/list/dovecot-cvs/2007-January/007385.html
The last patch is the most important one, if my guess is correct. You should then start seeing "Record field cache points outside record size". I'm still not sure why that happens though.
Hmm. Actually I think that would also explain the memory corruption. Great, one less thing to worry about. Now please just tell me that my fix works? ;)
I've been foraging among our backup indexes and finding tapes that haven't gone off site yet. I've had three cases from today of the "Corrupted Index file" but only two crashed (one with segfault, one with bus error). I'll need to pick up tapes tomorrow morning from our other on-site tape store to get the index files. Some of the folders themselves don't appear to have changed recently, but I'll get the backup copies of them too, to be sure. I think I'll have three cases to play with as a beginning.
If I'm right, and it's the processing of these index files causing the memory corruption, I should be able to reproduce it using the recovered files on my test server. I can then try your debugging patches and the latest fixes.
Best Wishes, Chris
-- --+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+- Christopher Wakelin, c.d.wakelin@reading.ac.uk IT Services Centre, The University of Reading, Tel: +44 (0)118 378 8439 Whiteknights, Reading, RG6 2AF, UK Fax: +44 (0)118 975 3094