[Bug]: Concurrent users broken with job-manager

[outlookhazy]

What happened?

When using the job-manager, concurrent requests from separate browsers/users seem to be causing a deadlock that makes the UI entirely unresponsive, even after all jobs are complete. Docker instance has to be rebooted to get it back.

Version

0.0.1 (Default)

What browsers are you seeing the problem on?

Chrome

Where are you running the webui?

Linux

Custom settings

No response

Relevant log output

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[outlookhazy]

Couldn't figure out where it's hanging, I tried swapping out the int list for a proper semaphore but python isn't really my thing
For clarity sake, forcing job-manager to disabled via source and it works fine with multiple client sessions (requests get queued by the system and it just shows Loading for longer while it's busy).

[outlookhazy]

Still present in latest commit (67515f0).
Steps to duplicate:
Build/run Docker container (I haven't tested with other setups)
Open 2 browser sessions on host directed to webui
pre-enter prompts into both sessions, no need to change any settings from default
Hit "Generate" on first session
while generation is in progress, hit "Generate" on second session
Both sessions will complete successfully and display the output image
Any attempt to perform any generation operation (or server request such as refresh etc) will now fail, this includes other tabs and any new sessions.
Websocket info in console shows rpc(?) queue actions going up, but I'm assuming something deadlocked in the background is preventing the system from recovering.

It appears necessary to restart the container to recover in this state.