It doesn't make sense for use to use env vars like "CONCURRENCY" to indicate the celery worker concurrency. To simplify, we will use the upstream defaults. The most controversial IMO of these is the Celery concurrency defaulting to number of cpus. I don't think this makes any sense, as a 64 core server running GlitchTip and many other apps doesn't likely need to have 64 celery workers.
This is a minor breaking change, so we should increment to 2.0 when this is released.
Smaller instances should just use the "good enough" defaults. For those customizing
- MAX_WORKERS > UWSGI_WORKERS
- MIN_WORKERS > UWSGI_CHEAPER
- INITIAL_WORKERS > UWSGI_CHEAPER_INITIAL
- CONCURRENCY > CELERY_WORKER_CONCURRENCY