On Fri, 2003-04-04 at 19:52, David E. Fox wrote: > > The most likely culprit is Mailman and Python, and memory is the > > quickest and easiest fix. I'd run top in a terminal, then sit and watch > > One thing is supposedly postfix is better at this than sendmail > is. There was one time I sent off a (non-spam!) mailing to 200+ > addresses, basically at once through a shell for loop. The box's load > average climed to over 20, and there was a whole lot of disk > activity. Postfix supposedly doesn't need to spawn so many copies, or > each fork takes fewer resources than in sendmail. Anyway, I wouldn't > probably do the same thing without inserting a 'sleep' in the middle > of the loop. >
Heh :-) A real good test of relative performance is to find a mail relay with a borked-up /etc/resolv.conf and about 10,000 messages in the queue. I've seen this situation with: (2) Sendmail Switch installs on Compaq Proliant 1650Rs on a DS-3 (1) Postfix install on a Dell PowerEdge 4350 on an OC-3 After fixing DNS and hup'ing the MTAs, Sendmail took five or ten minutes to clear the queue. Postfix took ninety seconds. Neither system was tuned. -- Jack Coates Monkeynoodle: A Scientific Venture... Are you compliant yet? ftp://ftp.rfc-editor.org/in-notes/rfc3514.txt
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
