Recently there has been an increase in devices that attach to World Community Grid and pull down the maximum number of allowed tasks (5 * number of processing cores). Some of these devices disappear quickly and never return any of the tasks they there were assigned.

Those assigned tasks time-out and are then assigned as high priority tasks to other users. Due to the restrictions on who can be assigned these high priority tasks and the rate at which they are being created we are having troubles distributing work efficiently.

In order to reduce this issue, we are instituting a new policy where newly connected devices will only be allowed to have a maximum of tasks in progress equal to the number of processing cores the devices has plus two more. This special limit will be in place for only 3 hours after the device is connected. Then the normal limits will apply.

As part of this we are currently experiencing issues with effectively distributing work. We have over-scheduled workunits that were going to timeout and set these tasks so that they are not assigned as high priority tasks. However, they are still associated with a platform (Windows, Linux, etc). Volunteers will see an unusually larger number of one of the following two messages over the next several hours while this backlog is reduced:

Message #1: Message from server: We are currently experiencing high load and are temporarily deferring your scheduler request. Your client will automatically try again later.

Message #2: Tasks are committed to other platforms

We apologize for these challenges as we adapt to the new participation patterns of some of our new volunteers. We expect this change to largely mitigate this situation.