I have multiple machines and they all have different settings as far as how many days worth of work they should download and queue up. But for a long time now, they never do. They always download the exact amount of work they can run and no more.

Is this a setting each project can set on their end? To override preferred 'days of work' settings for connecting clients and only give them as much as they need right now, and no more?

I ask because I do 4 main projects but having just added a 5th, suddenly that one for the first time in forever is downloading extra work for me. I'm just wondering what the deal is with that?

Some projects, in fact, most projects set a physical limit on how many tasks a host can hold or be sent. Projects set rules for how many cpu tasks or how many gpu tasks based on the number of cores in the host or the number of gpus the host has.

This does not always align with the days of work setting in the client.

My favorite project has a limit of sixteen gpu tasks per host. That's it. No more than that can ever be held on the host. Granted some tasks take half a day to run, but some only take an hour or less. You just have to accept what the project imposes for their rules.