Auteur: Rob Butterworth Date: À: exim-users Sujet: [Exim] exim process and database size problems (exim 3.12-10)
We're running exim 3.12-10 on Debian 2.2 with a 2.4 Linux kernel and xfs
file systems, and on one of our servers we're seeing some serious problems
with exim.
Mail tends to take longer to deliver that when we had sendmail, which users
find irritating. We have about 100 users on a 1GHz PIII system, with U160
SCSI disks and 512MB RAM.
The more serious problem is that the exim processes sometimes grow so large
that the system runs out of both physical and virtual memory. This may
correspond with the size of the database files in /var/spool/exim/db
getting *huge* - at one point the retry database was 5GB in size (which
wouldn't have been possible on an ext2fs partition....)
Has anyone else ever seen a similar problem ? Is anyone else running exim
with xfs as their filesystem ? I know it's pretty bleeding edge for a
production system, but we really want the journaling and quick recovery
times of xfs....
Rob
--
Rob Butterworth (rob@???)
Director of IT
Accelrys