All,
I have a customer who has over 100,000 subscribers to 35 or
so mailman mailing lists (ranging in size from 5 to 22,000
subscribers each). From time to time the CPU utilization goes to 100%
for exceptionally long times. Usually when I see that it has been
like that for several hours or a few days, I will manually kill and
restart mailman. Sometimes it goes right back to 100% CPU and some
times it will wait a little bit before returning to whatever it's
doing with all it's time. As of now it has been at 100% (mostly user
time) for over 12hrs. It seems that if the CPU was in a lot of iowait
then a restart fixes it, if it was user time then the CPU utilization
resumes after the restart.
The process that is using all this CPU time is the
BounceRunner. The system is a Celeron 1.7 w/1G RAM (I have a Cel 2.4
just laying around, I can upgrade easily, but is that enough ?). I
see from posts around the `Net that the bounce runner is a bit of a
pig. Is there any way to see the queue of work it has to do, is
there any way to determine if something is broken and causing the
high CPU utilization ??? I'd like to know if I need to upgrade the
server CPU, or need faster RAM or split the lists to different
servers, or if the work that is happening is just fine taking this
long to process. I figure the best way to start is to figure out what
work is taking so much CPU time. Any thoughts ? Any performance
tuning tips on the BounceRunner ?
RPM installation of Mailman on CentOS 3.5
mailman-2.1.5.1-25.rhel3.5 (version that comes with CentOS 3.5)
Postfix is the MTA
Thanks,
Mike
************************************************************
Michael J. McCafferty
Principal, Security Engineer
M5 Hosting
http://www.m5hosting.com
You can have your own custom Dedicated Server up and running today !
RedHat Enterprise, CentOS, Fedora, Debian, OpenBSD, FreeBSD, and more
************************************************************
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
