In a distributed GKE dask cluster, I have one graph that stalls with the traceback below. The worker dashboard continues reporting the same constant high value for cpu, whi
