Worker timeout on web UI for certain operations

[kjallen]

If I try certain operations via the web UI, for example regenerate spritesheet, with a set of images greater than 500 or so, I will get a timeout on the worker threads:

vikusdocker-1 | [...] [1] [CRITICAL] WORKER TIMEOUT (pid:10) vikusdocker-1 | [...] [1] [ERROR] Worker (pid:10) was sent SIGABRT! vikusdocker-1 | [...] [219] [INFO] Booting worker with pid: 219

The job will restart a 2nd time, starting with the same set of images and eventually find the same fate after about 60 seconds.

If I run via the cli tool, the same operation is able to complete, which leads me to believe something with the websocket/request layer.

I've tweaked the uvicorn (timeout_keep_alive) and FastAPI settings (timeout) and still see the same behavior.

Any suggestions on settings or changes to make to address?

[cpietsch]

Hey @kjallen, thanks for reporting the issue. Can you run the dev docker via docker-compose.dev.yml and run the server with log_level="debug" (python main.py) ? You should get more debug information - please post your logs here again. Can you also provide the Collection URL you are having problems with ?