Re: [Savannah-hackers-public] [Savannah-help-public] [gnu.org #705563] many messages missing from mail archives

[Ward Vandewege via RT]

Hi Karl et al,

This is going to require being very careful.

A few thoughts:

a) the spamassassin config in question applies to all gnu.org mail. If we are
going to make changes, we should likely make a separate config for lists.

b) the spamassassin config we run has exactly 2 score tweaks, neither of
which apply here. Local score tweaks are discouraged by the SA hackers.

c) the spamassasin scores we use are adjusted nightly from the
updates.spamassassin.org and sought.rules.yerp.org channels. In other words,
the weighing of the scores is adjusted continually, and is likely slightly
different now from what it was when the message below was processed.

Broadly speaking, every spam that makes it through to a mailing list is 
expensive
in terms of processing and archiving on lists, as well as wasted time by list
recipients.

Before we do anything to weaken the SA filter we have currently, I strongly
suggest we figure out how bad the false positive rate is, exactly. I really
do not want to go back to the bad old 'lists is sending lots of spam' days.

Karl, you said you think thousands of messages are missing; can you tell us
how you came to that conclusion? 

The obvious place to start looking for false positives would be in /spam/ on
lists, where the last 60 days of SA-tagged spam that reaches lists is kept.

Karl, you can access that directory now. Perhaps just looking through the
messages there for any messages with a 'From' address that also has messages
in the last few months of lists archives would be a good way to narrow down
the list of messages to look at more closely.

Thanks,
Ward.

-- 
Ward Vandewege <address@hidden>
Free Software Foundation - Senior Systems Administrator