Hmm, maybe that runs-every-60 seconds background query that took 27 seconds to run (!) was the source of our deadlock problems.. perhaps