Hi,
Jens Ahlin wrote:
I have upgraded our mailserver to Tao Linux from RH 8 and cyrus from 2.1.12-3 to cyrus 2.2.8-1 using Simons RPMS.
All works ok until I shutdown cyrus and starts it again. The the error below hit me in the face. To recover I delete the deliver.db and all under /var/lib/imap/db/. The it starts just fine, creates the files and runs happily utalizing the duplicate db.
I have built db4 (db4-4.1.25-8) with --disable-posixmutexes according to https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=91933.
Any ideas of how to solve this ?
I am now seeing the same issue under Red Hat Advanced Server 3.0 and Simon's RPMs.
In addition to the messages Jens mentioned I'm seeing the following on subsequent restarts:
Oct 12 13:14:18 lvr13 master[13034]: process started
Oct 12 13:14:18 lvr13 master[13035]: about to exec /usr/lib/cyrus-imapd/ctl_cyrusdb
Oct 12 13:14:18 lvr13 ctl_cyrusdb[13035]: DBERROR db4: malloc: Cannot allocate memory: 3435973880
Oct 12 13:14:18 lvr13 ctl_cyrusdb[13035]: DBERROR db4: PANIC: Cannot allocate memory
Oct 12 13:14:18 lvr13 ctl_cyrusdb[13035]: DBERROR: critical database situation
I don't really understand what's going on. We are using Intel Xeon processors that support NPTL. So I don't think that all the other issues with BDB4 that are mentioned in bug #91933 apply at all.
I suppose we could try rebuilding db4 without NPTL-support, but I'd really prefer to use the RPMs provided by Red Hat. The one we have is db4-4.1.25-8. Since not all apps that use BDB4 break, I guess cyrus-imapd must be doing something different. Does anybody have any insights regarding this? I've searched the list archive, but while there were many problems people had with BDB, none of them seemed to explain this particular situation.
Thanks, Sebastian Hagedorn -- Sebastian Hagedorn M.A. - RZKR-R1 (Gebäude 52), Zimmer 18 Zentrum für angewandte Informatik - Universitätsweiter Service RRZK Universität zu Köln / Cologne University - Tel. +49-221-478-5587
pgpGgsWwHThzW.pgp
Description: PGP signature
