For a couple of days I have been getting the message to the effect that ARP units are committed to other platforms. As my machines are Windows 7, that means that there are units available but not for Windows.

The inference here is that non-Windows machines have been downloading too many units to complete within the deadlines or, less likely, many errors are being made. Possibly both scenarios.

Downloading too many would be selfish and slow the project down.

Mike

Joining Unixchick to point out that Linux users aren't getting work either, and will sometimes see that message... That said, there don't seem to be [m]any retries at the moment; perhaps most of the last backlog got cleared out as a side-effect of having two successive "lack of new work" intervals in rapid succession (and seeing off generation 148!), but give it another day or two and that message will show up regularly again

By the way, one other possible reason for "other platforms" might be 32-bit versus 64-bit... I allow for both cases on my systems and I occasionally get a Linux 32-bit ARP1 task (usually as a retry), but I know other users deliberately block 32-bit applications because they are slower; if 32-bit users miss deadlines, the scope for retries may be restricted if that's the case (whatever the O/S).

I also agree about the need to do something regarding the excessive number of missed deadline tasks; however, I fear that "user education" won't work so it's likely to have to be a technical solution.

Cheers - Al.

P.S. -- Mike, I'm sure you weren't singling out Linux for blame , and there's anecdotal evidence that the proportion of Windows tasks missing deadlines is probably similar to (if not greater than) that for Linux and Darwin combined...

And for those interested in climate work while waiting for more work here, CPDN has a batch of tasks out now. (about 1745 according to the server status. This is only updated about every two hours so may be higher or lower depending on whether the batch had finished uploading when the server state updated. Batch size was 3,900 tasks. When these run out or get close to it there is another batch of the same size going out. (There are some 32bit compatibility files needed for these batches. Details are under Linux section of CPDN fora. (both batches are Linux only as will be the multi-core tasks coming out soon that require 26GB RAM/task.)

Edit: Tasks were still being added to the hopper. Latest server status is over 2,600

I'm curious about the message too.

| World Community Grid | No tasks are available for Africa Rainfall Project
| World Community Grid | Tasks are committed to other platforms

Is the tasks are committed message from ARP? or MCM ? I ask for both, so I'm guessing someone only running one or the other can answer that one.

I wish the log message was clearer.
I'm guessing it is just a "fake" message for I've got nothing for you, but it doesn't mean that the system has something for another system right at that moment, but maybe it had something for another system in the last "period of time" . Maybe it means I gave a different system a resend within the last 30 minutes ??

My vote is for poor message design.

This has been mentioned in various threads in various forums in the past, including recently in the Chat Room's Project Status thread on 2025-07-21 -- probably worth repeating some of that here...

To work out which project(s) it actually applies to, look for an application that is missing from the list of "No tasks are available for ..." -- if only ARP1 and MCM1 are selected by the user, it's MCM1 that has tasks committed elsewhere in this case.

So Mike's repeated sightings of the message a couple of days ago were probably at the time there was a huge quantity of MCM1 retries for Linux tasks (I got nearly 100 on one day!); that is far from typical, by the way.
I think that guess is right for when the "other platforms" case applies. I'd need to do a code dive to confirm it but I suspect that message is probably produced by checking a flag rather than a list of affected applications, and I suspect it would only appear once if tasks for both ARP1 and MCM1 were committed elsewhere (with no "no tasks" line for either application!)

If that is the case, I doubt the message could be made much clearer, confusing though it might be. (Compare with "Too Late" versus "Can't use" on those Darwin validation failures...)

Cheers - Al.