ERPNextCalc Redis RQ Short, Default & Long Queue Isolation

ERPNextCalc Queue Memory Allocation (512 MB to 1,024 MB)

Each short worker handles lightweight 30-second tasks like transactional emails or payment gateway callbacks and averages 512 MB RSS. Default workers process GL reposting at 768 MB, while long workers spike to 1,024 MB during bulk CSV imports.

We recommend running a minimum 2:2:1 ratio (2 short, 2 default, 1 long) on 16 GB nodes, reserving 3,584 MB total for asynchronous queues.

ERPNextCalc Redis Port Separation (:11000 vs :13000)

Frappe v15 requires two independent Redis instances: redis_queue on port 11000 without aggressive LRU key eviction, and redis_cache on port 13000 configured with allkeys-lru.

We advise allocating at least 512 MB maxmemory to redis_queue and disabling RDB snapshotting on redis_cache to prevent fork latency spikes.

ERPNextCalc Redis RQ Split in common_site_config.json

{
  "redis_cache": "redis://127.0.0.1:13000",
  "redis_queue": "redis://127.0.0.1:11000",
  "background_workers": 5,
  "rq_queue_split": { "short": 2, "default": 2, "long": 1 }
}

Explicitly budgeting Redis RQ pools in ERPNextCalc eliminates silent background bottlenecks across warehouse and accounting workflows.

Adjust the Short, Default, and Long sliders on ERPNextCalc to recalculate total memory headroom before deploying Supervisor.