On Fri, Jun 02, 2006 at 09:38:32AM -0700, List Mail User wrote: > >My guess is that these came in before any of razor, uribl, etc, got ahold of > >them. I just checked them all: [...] > Many of the name servers involved have been SBL listed for weeks. > Also his Bayes failed horribly on some of these and even auto-learned some [...]
It is strange that there are no BAYES_* hits on most of the messages, but auto-learning as ham isn't necessarily a problem. > There are more problems here than just being at the leading edge > of a spam-run (DNS and net test timeouts maybe?). There could be. You'll notice the above says "my guess is X" not "your problem is exactly X". :) From my quick 5 second look at the issue, the main difference between the original rule hits and the list I have were uribl and razor. That could be a simple thing like the messages weren't listed yet (the most common issue), or something deeper that needs to be looked at such as dns timeouts, internal/trusted_networks config, skip_rbl_checks, etc. -- Randomly Generated Tagline: "For just $19.95 a month, you too can test your applications over the Internet." - Hal Stern about AOL
pgpaxmr53uA8o.pgp
Description: PGP signature
