From: Roger <ro...@bl...> - 2004-04-30 12:57:19
|
Hi, I just did install assp on a server with quite high volume. It's running well, also there are some details to solve and the load on the server is very high: - To make some samples for the SPAM, if there are not false positive, I make a tail -f maillog | grep spam. My problem is, that if you have many files, like 20-70k, it isn't so easy to find the file in reasonable tome. grep has a problem on solaris, so I need a perl, but it's still not so fast. If in the maillog is also the file name, this would be easy e.g.: Apr-30-04 14:38:26 193.8.177.221 <al...@ka...> to: han...@te... Bayesian spam [Fw_dossiers_--63459.eml] - a detail: if the subject is empty, I get files with "--234234.eml", and they are difficult to delete on UNIX. Something like _--1234.eml would make deletion easier (One solution was: rm -- --31855.eml). - Is there some performance tuning I can do? Some additional ideas: I saw, that spam mails often are coming in multiple copies, because spamers send to more than one destination address mails. So maybe statistics about the number of mails with the same body could help to find spams, even they consist only binary or little information? Some stats about top spamed destination addresses would be also nice. I'm impressed about the simple installation and easy to use web interface of assp. Best regards, Roger |