This is a classic n8n queue-mode + ephemeral volume issue (very common on Railway).

Root cause:

Your workflows are queued, but there is no worker actually executing them OR

n8n is losing its state because the filesystem is not persistent.

Checklist to fix it:

1) Confirm execution mode

If you have:

EXECUTIONS_MODE=queue

then a worker service is mandatory.

Without n8n-worker, executions will stay in "queued" forever.

2) Make sure a worker is running

You must have a separate service/container with:

n8n worker

Environment variables must match the main service:

- QUEUE_BULL_REDIS_HOST

- QUEUE_BULL_REDIS_PORT

- QUEUE_BULL_REDIS_PASSWORD (if any)

3) Redis is REQUIRED

Queue mode does NOT work without Redis.

Postgres alone is not enough.

4) Persistent volume (this is why it keeps asking to mount)

Railway filesystem is ephemeral.

You MUST attach a persistent volume and mount it to:

/home/node/.n8n

Both the main n8n service AND the worker must use the same mounted volume.

5) Permissions check

Ensure the mounted volume is writable by the node user:

chown -R node:node /home/node/.n8n

Quick confirmation test:

Temporarily set:

EXECUTIONS_MODE=main

Restart the service.

If workflows execute immediately → the issue is 100% missing/misconfigured worker

or non-persistent volume.

Once worker + Redis + persistent volume are correct,

queued executions will start processing instantly.

Hope this helps — this setup fixes 99% of “queued forever” issues on Railway.