What version of client are you running?

If WCG had too large a share of the computing time, the client will stop fetching work until the other active projects have had their share, but what I found was that projects with short run times and short deadlines confuse the client scheduler so I've set switch times now from default 60 minutes to 35 minutes i.e. shorter than the shortest total run time of a task and the cache smaller that the shortest deadline, with is sometimes just 2 days.

With client version 6.10 you can go to the project tab, select e.g. WCG and hit the properties button. That will tell if WCG has a positive or negative Scheduling priority and same for CPU work fetch and deferral. If those are zero or negative, then no work will be requested.

Resetting WCG does set those values to zero or close to it, but if the other projects have positive values the client still will not go out.

How to force the issue: Suspend work fetch of your other projects and wait a bit or give WCG a bigger Resource Share/Project Weight. The default for all is 100. Set it to e.g. 500 and WCG gets more time allocated.

CPDN could hog your CPU but that would be indicated by a High Priority processing message. I only allow one task of them at the time, and low resource share so it gets about 1 hour a day, 300 hours projected as total run time. The task after 4 months has done 6 days [144 hours] at 38% progress and is due in December... no rush.

I have 16 CPDN wu's on a rig but it still pulls WCG work so doubt that's the issue.
Try suspending CPDN then try to pull down some WCG work.....then unsuspend cpdn at your leisure.