Hello all, My goal today is to get this filtering working on my 4 mail servers. Just a summary of my situation. I dont know if anyone is using this is a heavy production environment ( I assume so ) but I am running 4 Quad Xeon servers (1 Gig RAM) and Spamassassin (spamd) and routing mail that has a score of 10 or higher to dev/null.
This is working very well. We are a heavily spammed organization with a bit over 70,000 users. Without spamassassin on, my loads are normaly .5 - 2 (pop gets a bit slugish and stops responding after 4 or 5 ) With SA running the load is around 1.5 which is fine. I am running SA 2.43 from an RPM in totly stock config ( except for changing limit from 5 to 10 ) Now the problem: I am seeing huge CPU spikes over time. I can run for a few minutes to almost an hour and then the CPU will load up to 30.00 or more. Obviously this is killing the machine. So them I have to shut SA off and let it catch up. I see in the log files that during this time the spamd takes upwards of 30 seconds to flag a message. So that must be part of the problem. Or probably a symptom. In this environment ( 6000 - 15000 messages and hour ) should I be running the stock config or taking some of the checks out? Are some hanging or taking too long? I need to know what to check and how to determine the problem. I hope that I can give some feedback for this product in a high volume environment but also need to get this working so that it does not kill the machines..... Looking forward to any suggestions. Thanks Mark Quoting "Clayton, Nik [IT]" <[EMAIL PROTECTED]>: > > I am using spamc for all users through /etc/procmailrc on 4 > > servers and spamassassin works great. The problem is that it loads up > > the CPU to 4.0 and often 10.0 or above. How can I reduce this? > > Eliminate razor check is one I have seen on some other posts. What > > would be the best to try? > > [...] > > > Suggestions appreciated. > > > > I am running RedHat 8.0 on Quad Xeons with spamassasin (spamd) > > version spamassassin-2.31-16 ( version OK? ) > > with this for an /etc/procmailrc file: > > Are you actually seeing delays in the mail processing? > > I don't think you have a problem. A load of 4.0 on a 4 CPU box is > equivalent to a load of 1.0 on a 1 CPU box. > > With 4 CPUs, Sendmail's not going to start thinking the box is loaded > until you hit a load of (8 x 4) 32. > > You might want to make sure that all the temporary files that are being > created (note: not mail queue files, temporary files) are on a memory > backed filesystem, rather than actual disk, to drop the IO load. And > make sure that SA isn't checking large files (say, anything over 100K). > > But apart from that, it sounds like you've still got a fair amount of > oomph left in your servers. > > N > -- > 1 1 2 3 4 5 6 7 7 > 0 0 0 0 0 0 0 5 > -- The 75 column-ometer > Contributing to the heat death > Global Messaging, 120 Cheapside, x83331 of the universe since 1973. > ---------------------------------------------------- This mail sent with 3Web WebMail http://www.3web.com ------------------------------------------------------- This sf.net email is sponsored by: Influence the future of Java(TM) technology. Join the Java Community Process(SM) (JCP(SM)) program now. http://ads.sourceforge.net/cgi-bin/redirect.pl?sunm0003en _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
