Hello,

Serves become frequently unavailable recently. Since gpu wus take very little time to complete and number of wus per host is limited, isnt anyone concerned about the project's sustainability?

Regards.

I can confirm. The database outages occur regularly every day late morning (9-12 UTC). It looks like someone is working on the db server or there is a scheduled backup task that is taking too long.
Increasing the maximum number of tasks per host would really help with keeping our queue full. I might attach a backup project with share=0.

Project, users and the forum used to be very active in the days of radeon 5xxx series...