OK I am running 5.4 of the scheduler as that is what is available on the latest debian. Anyway, what I want to do is run my 3 projects differently then it looks like I can. I want to download from one project and complete that task, upload, then go to the next project, download, run, then upload. I don't see a way to set that up. All I am seeing is that it switches between projects every hours, or whatever interval I set. Is there a way to have it complete a WU on a project before going on to the next project? On my old slow PC that would be more effecient that switching so much.

Without a lot of micromanaging, not really. To do it manually, attach the projects and set all to No New Work. Let everything in your cache run out (each project will fetch upon attaching). When you've uploaded and reported everything, enable work fetch for one project only. Let it download. Set it back to No New Work. When that unit is finished and uploaded then enable work fetch for the next project, etc. This probably will become a large pain in the butt.

A very small connect interval and a very large (bigger than the longest workunit) switch interval might get close to what you want, but you may have work for more than one project on hand.

Are you leaving applications in memory while suspended? If you are, then there's no work lost on the switch between projects. You'd only lose work upon closing BOINC. But you'd still be losing that work even if you were only working on one project.

Is it possible for you to download the latest client from Berkeley (as I understand it, Debian is quite slow in updating their repositories)? The scheduler is extremely different and will only switch upon reaching a checkpoint. That way you wouldn't need to leave apps in memory (if memory is an issue for you). I have no idea how hard it is to set up BOINC under Linux from the Berkeley download. That's high on my list of things to do as soon as I get the Ubuntu iso to burn and then installed.

---

Kathryn :o)