Environment:

- Project: n8n (production)

- Services: Primary, Worker, PostgreSQL, Redis (based on n8n template)

- PostgreSQL Volume: geese-volume (contains production data)

Issue Summary:

Our n8n deployment has been experiencing critical PostgreSQL connection failures since attempting to resolve timeout issues. The PostgreSQL service now fails to authenticate with our existing data volume, causing a complete service outage.

Initial Problem:

- Error: "timeout exceeded when trying to connect"

- PostgreSQL checkpoint operations taking 228+ seconds

- n8n services showing "Error 503: Database is not ready"

Current Critical Issue:

After recreating PostgreSQL service, the existing volume (geese-volume) cannot authenticate with new service credentials: 2025-09-08 21:47:37.050 UTC [7266] FATAL: password authentication failed for user "railway" 2025-09-08 21:47:37.050 UTC [7266] DETAIL: Role "railway" does not exist.
Connection matched file "/var/lib/postgresql/data/pgdata/pg_hba.conf" line 128: "host all all all scram-sha-256"

Attempted Solutions:

1. Initial timeout fix attempt:

- Added PostgreSQL tuning variables (POSTGRES_MAX_CONNECTIONS, etc.) - caused deployment failure

- Removed all custom environment variables - PostgreSQL started but timeouts persisted

2. Service recreation attempts:

- Deleted and recreated PostgreSQL service

- Reattached the existing geese-volume at /var/lib/postgresql/data

- Result: Authentication failure - new service expects user "railway", old database doesn't have this user

3. Authentication bypass attempts:

- Added POSTGRES_HOST_AUTH_METHOD=trust - PostgreSQL fails to start entirely

4. Template redeployment:

- Deployed fresh n8n template

- Mounted existing geese-volume to new PostgreSQL service

- Same authentication mismatch issue persists

Critical Information:

- The geese-volume contains production data that must be preserved

- Original PostgreSQL service ID was: cb757d5e-2311-4f08-8f3a-7981ee22ab3c

- Volume mount shows in logs: /var/lib/containers/railwayapp/bind-mounts/[id]/

vol_amf28lpo4voq7qzc

- n8n services cannot start without database connection

Error Logs:

# PostgreSQL slow checkpoint (initial issue): 2025-09-08 17:17:51.447 UTC [73] LOG: checkpoint complete: wrote 66 buffers (0.4%); write=182.098 s, sync=228.824 s, total=697.970 s

# n8n timeout errors: Error: timeout exceeded when trying to connect at /usr/local/lib/node_modules/n8n/node_modules/pnpm/pg-pool@3.6.2_pg@8.12.0/node_modules/pg-pool/index.js:45:11

# Current authentication failure:

FATAL: password authentication failed for user "railway"

DETAIL: Role "railway" does not exist.

# n8n connection refused:

Initializing n8n process. There was an error initializing DB connect ECONNREFUSED fd12:549b:ec6:0:1000:6:92dc:b607:5432

Request:

We need assistance with:

1. Recovering access to the existing PostgreSQL data in geese-volume. The original Postgres modul was cb757d5e-2311-4f08-8f3a-7981ee22ab3c.

2. Understanding why PostgreSQL fails to start with POSTGRES_HOST_AUTH_METHOD=trust

3. Properly matching authentication credentials between the new PostgreSQL service and the existing database

4. Preventing future timeout issues once connection is restored

Impact:

The production n8n instance is completely down. All workflows and automation stopped. Critical business processes affected.

Thank you for your urgent assistance