https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6444
--- Comment #6 from Bradley Kieser <[email protected]> 2010-06-03 20:23:12 EDT --- (In reply to comment #4) > > Bradley, can you confirm the new patch diffed from the > > current version solves the issue and works for you, please? > > No need to, it _is_ the same thing, the diff is just another > representation of the proposed replacement module. > > I haven't tried it yet, but I have great expectations: I tried > twice in the past to switch bayes from MySQL to PostgreSQL > (different version), and every time it turned out that the > 'IN' operator yields very poor performance the way it is > used by SA. I hope to be able to try this new approach soon. > Other experience and review is most welcome. Hi Mark, This patch will certainly address that issue for you. There are a few more areas that I will be looking at next, the awl table inserts frequently kick out a duplicate key error (minor) and running sa-learn can kill the system too... again it looks like there is a nasty "IN" clause being used there too. For us, if we run sa-learn at the moment the exim processes quickly stack up (we use exim) in what looks like lock waits. Not fully investigated that yet, but thought that it was best to mention that here because I would hate for you to put the patch in only to be killed with sa-learn! -- Configure bugmail: https://issues.apache.org/SpamAssassin/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.
