Bitten by Berkeley

[Sebastian Hagedorn]

Hi,

just an anecdote:

Today we had to relocate our Cyrus 2.2.12 server to the other node in the 
failover domain for maintenance purposes. It was shut down cleanly and then 
restarted on the other node. The startup began with lots of messages like 
this one:

Mar 27 09:55:57 lvr14 master[26599]: about to exec 
/usr/lib/cyrus-imapd/ctl_cyrusdb
Mar 27 09:55:58 lvr14 ctl_cyrusdb[26599]: DBERROR db4: Log sequence error: 
page LSN 5525 10229151; previous LSN 5528 3779206
Mar 27 09:55:58 lvr14 ctl_cyrusdb[26599]: DBERROR db4: Log sequence error: 
page LSN 5525 10229151; previous LSN 5528 3779314
...

Then it seemed to start and work OK, but it got stuck as soon as the first 
ctl_cyrusdb was started. From then on it didn't accept any connections. I 
restarted the server, this time without any error messages. A little later 
we relocated the service back to the original node. Everything appeared 
fine initially, but then these familiar messages started showing up:

Mar 27 10:29:25 lvr13 pop3s[8167]: DBERROR db4: fatal region error 
detected; run recovery
Mar 27 10:29:25 lvr13 pop3s[8167]: DBERROR: error fetching 
8|3Í_^RESC^Uvüe^L^LÚO÷*û2: DB_RUNRECOVERY: Fatal error, run database 
recovery
...
Mar 27 10:29:25 lvr13 imaps[8820]: DBERROR db4: fatal region error 
detected; run recovery
Mar 27 10:29:25 lvr13 imaps[8820]: DBERROR: dbenv->open '/var/lib/imap/db' 
failed: DB_RUNRECOVERY: Fatal error, run database recovery
Mar 27 10:29:25 lvr13 imaps[8820]: DBERROR: init() on berkeley

I tried removing the databases and the environment (we have only used 
BerkeleyDB for deliver.db and tls_session.db anyway), but for whatever 
reason nothing seemed to work. Then I decided to do what I had planned for 
the 2.3 upgrade anyway: I moved those two databases to skiplist format 
(facilitated as always by Simon Matter's RPM). Now everything is fine again 
and will hopefully *stay* that way! Until now I had thought that the issues 
with BerkeleyDB were manageable, but I guess there just isn't any point.

Cheers, Sebastian Hagedorn
-- 
     .:.Sebastian Hagedorn - RZKR-R1 (Gebäude 52), Zimmer 18.:.
Zentrum für angewandte Informatik - Universitätsweiter Service RRZK
.:.Universität zu Köln / Cologne University - Tel. +49-221-478-5587.:.
                   .:.:.:.Skype: shagedorn.:.:.:.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 186 bytes
Desc: not available
Url : https://lists.andrew.cmu.edu/mailman/private/info-cyrus/attachments/20060327/28128cb0/attachment.bin