Is it possible that this might have been related to the stats being broken? The stats SQL was "stuck" and I had to rebuild it with repair table use_frm, bringing it's size down from over 10 gigs to about 400m too, and since then the clamd problem has not happened? |