Short answer: Queue mode separates the process that receives and coordinates work from workers that execute it. Redis carries the queue. Use it when execution volume or isolation requires more than one worker.

What changes in queue mode?

The main n8n process handles the editor and coordination. Workers pick up queued executions. Redis stores the queue state that connects those roles. Database access remains part of the deployment and must be shared correctly.

Queue mode adds operational parts. It does not fix a slow node, a poor database query, or an external API limit by itself.

Which values must match?

Use the same encryption key and database connection across the main process and workers. Use the same queue configuration and Redis endpoint. A mismatch can make a worker unable to decrypt credentials or find queued jobs.

Keep worker concurrency within the capacity of the database and the APIs called by your workflows. More workers can increase pressure rather than reduce it.

How should a rollout be tested?

Start Redis and the shared database first. Start the main process. Then start one worker and run a small workflow. Confirm that the execution is picked up and that the result appears in the editor.

Stop the worker during a test execution. Check how the job remains queued or is recovered. Restore the worker and verify that the execution completes without duplication.

What should be monitored?

Watch queue depth, worker health, execution duration, failed jobs, Redis memory, and database connections. A growing queue with healthy workers often points to a downstream limit or a workflow that takes longer than expected.

Review the execution architecture guide before changing concurrency. For the container layer, use the Docker Compose deployment guide.