Handling Worker Concurrency

Questions
1

So I’m trying to understand how n8n handles the maximum number of concurrent jobs while in queue mode.

I have a queue mode deployment on fly.io. I have the following env var set: 'N8N_CONCURRENCY_PRODUCTION_LIMIT = ‘36’ set on worker machines and the main machine.

For the command I use to run each worker, I have it as worker --concurrency=12 and can have up to 3 workers running at a time.

But from what I can tell, the n8n application never gets above 10 concurrent executions, now matter how many workers are active.

Is there something off about my setup? I’ve tried different permutations of different flags, env vars, etc., but it doesn’t seem to change the actual concurrency.

2

The settings n8n is actually using might not match what you are expecting. There are a few debug level log messages in the concurrency-control service, so it might be a good first step to enable debug level logging and watch to see what’s really happening.