occasional bayes failure

[Glenn Steen]

2008/5/19 Greg Matthews <gmatt at nerc.ac.uk>:
> Glenn Steen wrote:
>>
>> Are all the relevant ones using cached results? If so, SA is never run
>> on them... That is, after all, the whole point of the SA result
>> cache:-).
>> Not sure if that could have something to do with it. Perhaps worth
>> checking though.
>
> uh... weirder...
>
> actually since fixing the Bayes table, the only ones without a Bayes score
> are indeed cached (good call). However... on the other two relays, cached
> hits are /still/ showing a Bayes score, here is an example:
>
> May 19 16:29:08 mailr-k MailScanner[7257]: Message m4JFT1Jb009726 from
> 218.37.8.84 (signs7 at 4hisgloryquartet.com) to nerc.ac.uk is spam,
> SpamAssassin (cached, score=23.389, required 5, autolearn=spam, BAYES_50
> 0.00, BOTNET 1.00, DCC_CHECK 1.70, DIGEST_MULTIPLE 0.00,
> RAZOR2_CF_RANGE_51_100 0.50, RAZOR2_CF_RANGE_E4_51_100 1.50,
> RAZOR2_CF_RANGE_E8_51_100 1.50, RAZOR2_CHECK 0.50, RCVD_IN_BL_SPAMCOP_NET
> 1.96, RCVD_IN_SORBS_WEB 0.62, RDNS_NONE 0.10, SARE_OEM_A_1 2.00,
> SARE_OEM_A_2 1.50, STOX_REPLY_TYPE 0.00, TVD_FINGER_02 2.13, URIBL_AB_SURBL
> 1.86, URIBL_BLACK 1.96, URIBL_JP_SURBL 1.50, URIBL_OB_SURBL 1.50,
> URIBL_RHS_DOB 1.08, URIBL_SC_SURBL 0.47)
>
> so which is right?
>
> Or perhaps the ones that hit the cache which dont have a Bayes score were
> cached when bayes was still broken... I'll zap the cache and see what
> happens.
Yep, this is what I'm thinking. Easy solution... remove the SA cache
DB... Perhaps do it like "service MailScanner stop; rm
/path/so/SpamAssassin.cache.db;service MailScanner Start" (don't
remember neither name nor path:-)... Don't remeber if that is needed,
but ... better safe thansorry:)
MS will recreate it.

> thanks Glenn
Glad to help.

> GREG
>

Cheers
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se