We have a sunblade 1000 running solaris 8 and a later version of sendmail.
Some users, kick off procmail to run spamc (which runs spamd/spamassassin)
and/or to do some filtering. This box gets 1.3 million messages/month,
with a total size of about 28Gig so it's not a huge amount of mail but
certainly not insignificant.
Periodically, we'll get a mail storm and one user in particular kicks off
a whole mess of procmail processes. Since sendmail is tuned to throttle
and back off based on the load, sendmail doesn't mash the machine but these
processes do. This user had spamc in his procmailrc above filters that
would have handled these messages and as soon as I changed the order of
the filters, things calmed down.
We're theorizing that the host was running out of processes and couldn't
fork spamc hanging these procmail processes. We therefore presume that
1/ If we pay attention to the order of rules we'll be better and
2/ If we increase number of processes (I think nproc in /etc/system(?))
this problem wont bother us again.
Anyone have any comments or seen something similar? Or know of a way to
limit number of procmail processes a user can run?
TIA
=-=-=-=-=-=-=-=-=-=-=-=- generated by /dev/dave -=-=-=-=-=-=-=-=-=-=-=-=-=-=
David Stern University of Maryland
Institute for Advanced Computer Studies
_______________________________________________
procmail mailing list
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail