On 2023-01-08 22:02, joe a wrote:
On 1/8/2023 3:18 PM, Carlos E. R. wrote:
On 2023-01-08 20:41, joe a wrote:
...
So I do not know.
Posting that on the SA list. I find it works the same with or without "score", so will go with that as it makes lint happy. Odd that lint complains, yet SA does not.
Ok, when you get an answer there, please tell here :-) On the other hand, I'm processing 1003 of the same mails (including 3 that are spam), using dnsmasq and bind, with the no forwarding trick, and I don't get the "open resolver" error from SA. But I see instead hundreds of errors in the machine running bind, it can not resolve many domains, others timeout. The original run did 81376 mails in 2217m21,9s, thus 36.7 mails per minute. Now I did 1003 mails in 1521 seconds, thus 39.6 mails per minute. Just a little bit faster (but might be because the procmail test file is much shorter). Now trying with: CVD_IN_ZEN 0 RCVD_IN_XBL 0 RCVD_IN_PBL 0 in .spamassassin/user_prefs [...] Ok, 1528S. No difference. 39.6 mails/minute. Then I looked into the mail log: <2.6> 2023-01-08T22:54:36.237088+01:00 Telcontar spamd 17988 - - config: failed to parse line, skipping, in "/home/cer/.spamassassin/user_prefs": RCVD_IN_ZEN 0 <2.6> 2023-01-08T22:54:36.237228+01:00 Telcontar spamd 17988 - - config: failed to parse line, skipping, in "/home/cer/.spamassassin/user_prefs": RCVD_IN_XBL 0 <2.6> 2023-01-08T22:54:36.237356+01:00 Telcontar spamd 17988 - - config: failed to parse line, skipping, in "/home/cer/.spamassassin/user_prefs": RCVD_IN_PBL 0 So that method does not work. And nullifying the results is pointless, too. [...] DOH! The documentation says: Setting a rule's score to 0 will disable that rule from running. https://spamassassin.apache.org/full/3.1.x/doc/Mail_SpamAssassin_Conf.html So it has to be: score RCVD_IN_ZEN 0 score RCVD_IN_XBL 0 score RCVD_IN_PBL 0 Testing again. 1512s, so 39.8 mails per minute. Barely an effect. Ok, let's kill all remote tests. skip_rbl_checks 1 Now it is 1334s, so 45.1 mails per minute. Well, that's faster, but not a lot of improvement. The next step is parallelize. It is easy to do with postfix, but in my case I'm post processing and not using postfix. But in fact, I can skip the spam checking and do only the sorting part, but that's my particular situation, doesn't help you. I was just doing tests of SA. -- Cheers / Saludos, Carlos E. R. (from 15.4 x86_64 at Telcontar)