Hello Filippo, Monday, October 23, 2006, 5:18:02 AM, you wrote:
> Hello Pete, since friday our mail server is overwhelmed by a very lot of spam > messages. Because of this the spool of my IMail Server gets full and > it actually get stuck. > Do you have any hint that can help me to fix this problem? There are a number of tricks to tuning IMail/Declude setups (I'm guessing from other posts that this is what you have). Using the AVAFTERJM option in Declude reduces system loads by only scanning messages for viruses after they have passed all of the spam tests. Since spam can easily be 90% of traffic these days this one option can save quite a bit of CPU for other tests. You will have to be careful to scan anything you release from quarantine for viruses however. ____ Through enlightened experimentation I have determined that low numbers in queue manager provide much better throughput. I have an IMail server that I use to process inbound spam and to test SNF. This single p4/2.4G CPU consistently handles 10 messages per second on average. By pushing this box to the edge (frequently) I have learned a few things about tuning it. My queue manger settings are: Listening Threads: 4 Retry Threads: 5 Delivery Threads: 8 Your mileage may vary!! -- The reason small numbers may be better than large ones is that your CPU(s) can really only process a handfull (about 2 per CPU on average) of threads concurrently. Any additional threads must wait and the OS must schedule them and resolve resource conflicts etc... That amounts to extra work. Keeping the number of threads small reduces overhead and allows the threads that are running to get more done. ____ One of our early boxes (now defunct) used Declude/Imail/SNF on NT4 - it was purposefully underpowered. On that box we discovered that running a local copy of Bind as a resolver and making 127.0.0.1 our primary DNS server improved performance quite a bit. Along these lines, be sure that long-running DNS queries are removed--- that is, if you have a DNS based test that takes a while to return then you're probably better off without it. ____ Hope this helps, _M -- Pete McNeil Chief Scientist, Arm Research Labs, LLC. ############################################################# This message is sent to you because you are subscribed to the mailing list <[email protected]>. To unsubscribe, E-mail to: <[EMAIL PROTECTED]> To switch to the DIGEST mode, E-mail to <[EMAIL PROTECTED]> To switch to the INDEX mode, E-mail to <[EMAIL PROTECTED]> Send administrative queries to <[EMAIL PROTECTED]>
