Dashboard not loading with Cythonized Scheduler

This appears to have started recently. Possibly related to PR ( https://github.com/dask/distributed/pull/4651 ) ( cc @crusaderky )? In particular loading the dashboard shows a 500 error. Locally the following exception comes up when the Scheduler is Cythonized. The issue does not occur when pure Python is used.

cc @quasiben

What happened:

distributed.utils - ERROR - object.__new__(distributed.scheduler.MemoryState) is not safe, use distributed.scheduler.MemoryState.__new__()
Traceback (most recent call last):
  File "/Users/jkirkham/Developer/distributed/distributed/utils.py", line 668, in log_errors
    yield
  File "/Users/jkirkham/Developer/distributed/distributed/dashboard/components/scheduler.py", line 296, in update
    meminfo = self.scheduler.memory
  File "distributed/scheduler.py", line 1913, in distributed.scheduler.SchedulerState.memory.__get__
  File "distributed/scheduler.py", line 330, in distributed.scheduler.MemoryState.sum
TypeError: object.__new__(distributed.scheduler.MemoryState) is not safe, use distributed.scheduler.MemoryState.__new__()

What you expected to happen:

Minimal Complete Verifiable Example:

import webbrowser
from dask.distributed import Client

client = Client()

webbrowser.open(client.cluster.dashboard_link)

Anything else we need to know?:

Environment:

• Dask version: 2021.04.1
• Python version: 3.8.8
• Operating System: macOS
• Install method (conda, pip, source): Conda

[jrbourbeau]

Alright, I tried in a fresh env and was able to reproduce the 500 error

[jakirkham]

Fixed with PR ( https://github.com/dask/distributed/pull/4761 ). Also CI is now covering Cythonization correctly ( https://github.com/dask/distributed/pull/4764 ). Closing this out