Try hitting "update".

Cheers

Windows host: 8 threads: 16 work units in progress Limit 20
Linux Host #1: 24 threads: 20 work units Limit 32
Linux Host #2: 32 threads: 6 work units Limit 32

Windows seems to be getting "enough" work units for the moment. The Linux machines are woefully lacking.

Cheers

I do run a zero cache on my Windows host that is getting no WCG work units, because not doing so caused problems with work units that have really short deadlines (the host is switched off overnight). However, I tried setting "No New Tasks" for all other projects and waiting until it was hungry for work units before hitting "Update" multiple times, but got nothing from WCG.

I have operated with a zero cache for a very, very long time, and its only WCG that is failing to send work units, and only recently (edit: and only on Windows). The message in the log is "Scheduler request completed: got 0 new tasks", not "Not requesting tasks: don't need (CPU: job cache full; Intel GPU: )".

Never knew WU were assigned by platform or platform dependent. Interesting. Disheartening as I only use linux for efficiency reasons, but interesting at least.

I don't think they assign WUs by/to platforms as such. They leave that to us. I remember reading here that certain projects performed better on different platforms.

If I read my notes, I see that I limited MCM to 20 on my most powerful machines which tells me that MCM was better on Windows and SCC better on Linux. But since MCM is all we get now all limits are off everywhere.