On Donnerstag, 17. April 2008 Tom Lane wrote:
> This latest bit with many fewer index entries than table entries is
> just weird, though. The 8.1.10 VACUUM bug could have led to having
> *more* index entries than table entries, but not the other way
> around.
Somehow I can think of it like that: SA didn't find user "vscan"
anymore, so created a new entry for it. And then it created all the new
learned ham/spam tokens again despite they were there already. Must
have to do with not finding it from the index or so.
It would be nice if pg_dump would FAIL on such a broken database. Like
this, I would have become informed about the problem. The bayes db
isn't quite interactive, and SA seems to ignore errors. The reason I
found it was that our SPAM filter started to let spam through which did
not happen for years (yes, we have *very hard* anti spam settings, our
customers love it that way - no spam). So I got aware that bayes got
stupid and then I found the mess...
mfg zmi
--
// Michael Monnerie, Ing.BSc ----- http://it-management.at
// Tel: 0676/846 914 666 .network.your.ideas.
// PGP Key: "curl -s http://zmi.at/zmi.asc | gpg --import"
// Fingerprint: AC19 F9D5 36ED CD8A EF38 500E CE14 91F7 1C12 09B4
// Keyserver: www.keyserver.net Key-ID: 1C1209B4
