We have tested N8N locally and are now looking to deploy it to Staging for further development and testing before moving to production.
Reading the process behind deploying on Heroku for N8N was similar to our on-premise deployment of Retool. Requiring us to create a Heroku App that uses the
container runtime so that the Docker Image is built on deploy.
Thus, we setup a Heroku container to try and deploy the N8N Latest (0.232.1) release. We were able to configure the heroku app to have all the
env variables, postgres and papertrail.
We forked and began to use: GitHub - n8n-io/n8n-heroku
We pushed up the git repo for a deployment, image built things were looking good. But inside of Papertrail post docker-build we can see the following inside error. In which N8N is not responsive at the associated Url provided by Heroku. It throws a 500.
Jun 15 13:17:18 [n8n-staging] Build succeeded
Jun 15 13:17:25 [n8n-staging] Starting process with command
Jun 15 13:17:26 [n8n-staging] N8N will start on ‘21821’
Jun 15 13:17:26 [n8n-staging] postgres://:@ec2.compute-1.amazonaws.com:5432/**
Jun 15 13:17:28 [n8n-staging] License manager not initialized
Jun 15 13:17:28 [n8n-staging] UserSettings were generated and saved to: /home/node/packages/cli/.n8n/config
Jun 15 13:17:30 [n8n-staging] State changed from starting to crashed
Jun 15 13:17:30 [n8n-staging] Process exited with status 1
Jun 15 13:18:12 [n8n-staging] at=error code=H10 desc=“App crashed” method=GET path=“/” host=n8n-staging-c2d46655f640.herokuapp.com dyno= connect= service= status=503 bytes= protocol=https
Have tried rolling back to different versions of N8N. Unable to find a version that works.
- n8n version: 0.232.1
- Database (default: SQLite): Postgres 15.3
- n8n EXECUTIONS_PROCESS setting (default: own, main): runs /entrypoint.sh from Web Dyno
- Running n8n via (Docker, npm, n8n cloud, desktop app): Docker / Container deploy on Heroku
- Operating system:
- Can anyone confirm they are able to run N8N on Heroku on the latest Postgres and Release? I see lots of support issues from years ago with folks struggling with deploying on Heroku. Curious if anyone can confirm this deploy is working (had past projects that failed to maintain their Heroku deployment) and want to de-risk our decision. If we can’t get this deployed we will have to move on.
- Can anyone provide next steps on how to best triage and understand why the App is Crashing? My prediction potentially is that the
PORTisn’t getting bound, I understand that Heroku has to handle a proxy between port 443 on the load-balancer back to N8N as indicated above it denotes which port it starts on.
Appreciate any help here.