SA timeouts after upgrade to 2.63

[Pete]

Need to search the archive on bayes, it gets discussed a lot and i am no
expert on it - have turned it on, but my low spec machines get crippled
by it.
Another reasons its off its fair administritive overhead and i am now
catching heaps of spam again - but this has taken a good 2 weeks to
settle down - the big tip was to reduce the max childs in ms.conf, this
helped me heaps.

Now i have learned how to add sbl_xbl.spamhaus i catch HEAPS more spam,
i have given this BL a score of 4.0 and spamcop 4 and my low spam score
is 5 so now back to catching loads of spam, Julian suggests a score of
6, but i think you will need bayes to get to there.

Turn bayes on and test, it wont work till it has seen 200 hams and 200
spams, then you will start seeing a score - you can set it to learn, but
not be used for scoring too...

>>
>>
>>
>
>Sorry, when I said "network monitor" I meant a general term for packet
>sniffer - for this kind of thing, the handiest I have found is to use
>'ngrep' - basically implements grep on the network - unbelievable useful -
>google it and you'll find it - you can just leave it running, piping to a
>file and then come back later and see what it found. Doing this with most
>other packet sniffers is not workable as they hold the captured packets in
>memory and cannot just be left running for long periods - also the ability
>to grep through the stuff and dump only the matches is incredibly useful.
>
>Anyway, sounds like a good plan on building a new server - I suggest you
>consider running BLs in MailScanner, though - that way you can just look
>at the MailScanner logs to see if one or more of them are timing out
>without having to run in debug mode.
>
>I'm unfortunately still getting alot of spam making it through - I was
>hoping that SA 2.63 would cut it down more. I'm not running Bayes, though
>- do you have any experience with that? Just wondering if bayes would help
>and how much.
>
>
>Goods luck,
>
>Ron
>
>
>
>
>
>