# Process default queue with local Raypythonmanage.pydjango_ray_worker--queue=default--local
# Process multiple queuespythonmanage.pydjango_ray_worker--queue=default,high-priority--local
# Process all queues with high concurrencypythonmanage.pydjango_ray_worker--all-queues--local--concurrency=50# Connect to Ray clusterpythonmanage.pydjango_ray_worker--queue=default--cluster=ray://ray-head:10001
# Sync mode for testingpythonmanage.pydjango_ray_worker--queue=default--sync
# Verbose outputpythonmanage.pydjango_ray_worker--queue=default--local-v2
The management command itself expects CLI flags and Django settings. The Docker entrypoint included
in this repository maps these environment variables into command-line options:
Variable
Description
DJANGO_RAY_QUEUE
Queue name passed as --queue
DJANGO_RAY_QUEUES
Comma-separated queue names passed as --queue; takes precedence over DJANGO_RAY_QUEUE
DJANGO_RAY_CONCURRENCY
Concurrency limit passed as --concurrency
RAY_ADDRESS
Ray cluster address used by sample settings and worker-cluster entrypoint mode