At 12:56 PM 4/2/2002 -0500, you wrote: >It's almost certainly a rule with a "bad" regex -- could you send me an >email with attached a sample of an email that causes the high cpu usage?
Once it starts -- shortly after I put up the CVS version almost every piece of mail seems to get stuck. I had ~50 spamd/spamc procs each of which used about 17mb or RAM. Things then get quite slow as the machine swaps things in/out. There is normally plenty of free RAM. After killing off sendmail I found that the messages eventually cleaned themselves up, but many took over 900 seconds -- with *no* network lookups (-L). Checks of what was printed from the debug flag revealed most of the stuck messages were not spam. The custom rules in local.cf work fine with 2.11 -- same local.cf file. Maybe I missed something. Is the AWL stuff turned on by default now? We're not using that. Is there something else I overlooked? If you have an idea how I can figure out exactly which message is causing problems I'd be happy to turn the box back into a guinea pig. We're calling spamc/spamd from within Procmail... reb _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
