There are reports of not enough available WUs.

I believe this is caused by the "accelerated retries" mechanism. So I have temporarily disabled this feature.

This mechanism normally works very well, but I think the current Pentathlon is pushing it to the limit. In particular, if too many WUs are being aborted, then the work queue is flooded with "failed" jobs that now have a higher priority, and these jobs require a "reliable host". To make things worse, any host aborting too many jobs is flagged as unreliable, so won't be given any of these tasks.

More...