[FIXED] Heroku - Dash: Port already in use

Issue

I wanted to deploy my plotly dashboard on Heroku, however upon running I keep getting information that the Port [PORT_NUMBER] is already in use. The port number is given by Heroku itself and changes everytime I restart the app. I tried setting up an app with a fixed port however it just stated that the given port is also 'in use by another program'. There are a few issues regarding something similar but they all pertain deploymend of node.js app and seemed to have been resolved by using a solution specific to node.js deployement.

Since I tried to locate the problem, I reduced my main module to minimal working example in app.py file:

from dash import Dash, html

app = Dash(__name__)
server = app.server
app.layout = html.Div("TESTING")
app.run_server(debug=False)

Procfile defined as follows:

web: gunicorn app:server

And here is the log from the running server I keep getting:

2022-06-05T15:38:23.696572+00:00 app[web.1]: * Serving Flask app 'app' (lazy loading)
2022-06-05T15:38:23.696693+00:00 app[web.1]: * Environment: production
2022-06-05T15:38:23.696830+00:00 app[web.1]: WARNING: This is a development server. Do not use it in a production deployment.
2022-06-05T15:38:23.696863+00:00 app[web.1]: Use a production WSGI server instead.
2022-06-05T15:38:23.696883+00:00 app[web.1]: * Debug mode: off
2022-06-05T15:38:23.702649+00:00 app[web.1]: Address already in use
2022-06-05T15:38:23.702652+00:00 app[web.1]: Port 45013 is in use by another program. Either identify and stop that program, or start the server with a different port.
2022-06-05T15:38:23.702869+00:00 app[web.1]: [2022-06-05 15:38:23 +0000] [489] [INFO] Worker exiting (pid: 489)
2022-06-05T15:38:23.850179+00:00 app[web.1]: [2022-06-05 15:38:23 +0000] [491] [INFO] Booting worker with pid: 491
2022-06-05T15:38:24.280232+00:00 app[web.1]: Dash is running on http://127.0.0.1:45013/
2022-06-05T15:38:24.280241+00:00 app[web.1]:
2022-06-05T15:38:24.280618+00:00 app[web.1]: * Serving Flask app 'app' (lazy loading)
2022-06-05T15:38:24.284103+00:00 app[web.1]: * Environment: production
2022-06-05T15:38:24.284156+00:00 app[web.1]: WARNING: This is a development server. Do not use it in a production deployment.
2022-06-05T15:38:24.284186+00:00 app[web.1]: Use a production WSGI server instead.
2022-06-05T15:38:24.284209+00:00 app[web.1]: * Debug mode: off
2022-06-05T15:38:24.295838+00:00 app[web.1]: Address already in use
2022-06-05T15:38:24.295891+00:00 app[web.1]: Port 45013 is in use by another program. Either identify and stop that program, or start the server with a different port.
2022-06-05T15:38:24.296221+00:00 app[web.1]: [2022-06-05 15:38:24 +0000] [490] [INFO] Worker exiting (pid: 490)
2022-06-05T15:38:24.702612+00:00 app[web.1]: [2022-06-05 15:38:24 +0000] [492] [INFO] Booting worker with pid: 492
2022-06-05T15:38:24.826989+00:00 app[web.1]: Dash is running on http://127.0.0.1:45013/
2022-06-05T15:38:24.826991+00:00 app[web.1]:
2022-06-05T15:38:24.827286+00:00 app[web.1]: * Serving Flask app 'app' (lazy loading)
2022-06-05T15:38:24.832077+00:00 app[web.1]: * Environment: production
2022-06-05T15:38:24.832127+00:00 app[web.1]: WARNING: This is a development server. Do not use it in a production deployment.
2022-06-05T15:38:24.832158+00:00 app[web.1]: Use a production WSGI server instead.
2022-06-05T15:38:24.832182+00:00 app[web.1]: * Debug mode: off
2022-06-05T15:38:24.834227+00:00 app[web.1]: Address already in use
2022-06-05T15:38:24.834256+00:00 app[web.1]: Port 45013 is in use by another program. Either identify and stop that program, or start the server with a different port.
2022-06-05T15:38:24.834479+00:00 app[web.1]: [2022-06-05 15:38:24 +0000] [491] [INFO] Worker exiting (pid: 491)
2022-06-05T15:38:25.150842+00:00 app[web.1]: [2022-06-05 15:38:25 +0000] [493] [INFO] Booting worker with pid: 493

It just keeps starting up and failing due to busy port.

What am I doing wrong?

Edit

Redeploying app from scratch solved the issue.

Solution

If the port is changing and the result keep being the same. There is following ways to try:

1. Verify that you are not running several servers on one port. For example, in your running configuration you are starting your server directly AND also using gunicorn simultaneously.
2. Try to redeploy the app from scratch. (Maybe it's some strange Heroku bug triggered on this specific deployment.)

Answered By - Nick Veld