Bryan<br><br>don't get hung up on load av - that's just a measure of how many processes are waiting for resource (cpu/memory/disk etc) and not necessarily an indication of 'slowness'. I've seen machines with load av well over 100 and running very nicely thankyou.<br>
<br>you can turn off the spamassassin checks in mailscanner.conf. also near identical machines will need different numbers for batch size and children size. <br><br>Also make sure you read the wiki on performance tuning etc.<br>
<br>Martin<br><br><div class="gmail_quote">On 20 April 2010 16:41, Bryan Guest <span dir="ltr"><<a href="mailto:bryan.guest@bmts.com">bryan.guest@bmts.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
Hi:<br>
<br>
This is a follow up to a previous post about the resident size of MailScanner.<br>
<br>
I have added a new incoming MailScanner blade to my setup. It's a Sun X2200 with 4Gb of RAM. My previous two blades are Sun V20Z machines also with 4Gb of RAM. All Machines are running Redhat ESv4<br>
<br>
I am running the latest Mailscanner and ClamAV-SA package on the new machine. The resident size of MailScanner is 170Meg on the new machine. (It's 76M on the best running machine).<br>
<br>
I am interested in how I can pair down the resident memory size used by MailScanner or improve performance/reduce load average.<br>
<br>
My fundamental problem is that even though the two V20Z's are running the same older version of MS (4.56.8), one runs hotter (higher load average) then the other. And the newer X2200 machine runs hotter still.<br>
<br>
IE: the one V20Z is at load average 0.5. The second is at 1.32. The X2200 with the latest MS is at 3.57. Note that these are in round-robin rotation behind an L4 switch.<br>
<br>
I have adjusted the number of processes as best I can to avoid swapping out based on the resident size of MS. But randomly it seems the load average will shoot up on the second V20Z or the X2200. And by shoot up, I mean that it goes over 25 (~30-50 sometimes) and sendmail stops accepting mail.<br>
<br>
Can anyone assist me in shedding any light on why one blade of two identical boxes runs hotter, and why a newer machine running the latest MailScanner runs even higher.<br>
<br>
Alternately I would be supremely receptive to tips on how to bring down the load on these servers. Note, I am running the log and spool partitions with noatime, and syslog with - on the logs.<br>
<br>
For clarification, we are not using SpamAssassin or the Bayes functionality on these blades. We employ MailScanner strictly for policy enforcement and virus scanning and use an external spam filter.<br>
<br>
If I install a fresh Mailscanner install, and the ClamAV/SA package but halt the install before it compiles SA, will this reduced the resident size of MailScanner?<br>
<br>
All the best regards to everyone, and thanks in advance,<br>
Bryan<br><font color="#888888">
<br>
<br>
<br>
<br>
-- <br>
MailScanner mailing list<br>
<a href="mailto:mailscanner@lists.mailscanner.info" target="_blank">mailscanner@lists.mailscanner.info</a><br>
<a href="http://lists.mailscanner.info/mailman/listinfo/mailscanner" target="_blank">http://lists.mailscanner.info/mailman/listinfo/mailscanner</a><br>
<br>
Before posting, read <a href="http://wiki.mailscanner.info/posting" target="_blank">http://wiki.mailscanner.info/posting</a><br>
<br>
Support MailScanner development - buy the book off the website! </font></blockquote></div><br><br clear="all"><br>-- <br>Martin Hepworth<br>Oxford, UK<br>