> On Tue, 2009-03-24 at 08:10 -0500, Bowie Bailey wrote: >> >> Your assessment sounds right to me. I would make two suggestions. >> >> 1) Memory is cheap these days. Add some more RAM. > > That's a mitigation strategy, yes, but it doesn't really answer OP's > question about how to make spamd stop trying to allocate new incoming > spams to try to process them all at the time they come in and instead > put them into a queue, in a effort to try to "even" the load out. > >> 2) Reduce the maximum children setting so that the system doesn't start >> swapping. This will cause SA to scan faster and should result in fewer >> messages slipping through while SA is busy. > > But it also means if the incoming load temporarily overruns the > available children currently available, then the excess doesn't get > spamd treatment. Or does it? > > If I have 5 spamd children available and (just to torture it) I fire off > 50 spamc processes, what happens? > > b. > Match your MTA processes to the spamd children. Your MTA will send 4xx 'busy now, come back to play later' message. Let the sending MTA queue it back up (or zombies will just go away)
-- Michael Scheidell, CTO >|SECNAP Network Security Finalist 2009 Network Products Guide Hot Companies FreeBSD SpamAssassin Ports maintainer _________________________________________________________________________ This email has been scanned and certified safe by SpammerTrap(r). For Information please see http://www.secnap.com/products/spammertrap/ _________________________________________________________________________
