Uploaded new Flask code, socket file disappeared, how do I bring it back?

January 29, 2017 303 views
Python Nginx Ubuntu 16.04

I start this off with the fact that I am a python developer and I know very little about server management so this is all a learning curve for me.

I created a new Ubuntu 16.04 droplet and I followed this tutorial - https://www.digitalocean.com/community/tutorials/how-to-serve-flask-applications-with-gunicorn-and-nginx-on-ubuntu-16-04. All was well and it worked.

I used Filezilla to move my flask application from my localhost to the droplet. I overwrote the old app.py with my app.py. wsgi.py was unchanged nothing else should have changed.

I ran sudo systemctl restart myproject and went to view the server from my browser and I received a 502 error.

Looking at the error log I have this:

2017/01/29 23:21:37 [crit] 18456#18456: *1 connect() to unix:/home/user/myproj/proj.sock failed (2: No such file or directory) while connecting to upstream, client: xx.xx.xx.xx, server: xx.xx.xx.xx, request: "GET / HTTP/1.1", upstream: "http://unix:/home/user/myproj/proj.sock:/", host: "xx.xx.xx.xx"

Which suggests there's no socket file. I ran ls -al in /home/user/myproj and I can verify it's not there. I didn't delete it so I am not sure what happened to it and how to make it return.

I've verified the nginx.conf in /sites-enabled for this project and in the systemctl .service and all the info matches up.

I tried restarting/reloading my project.service, nginx and even a reboot of the server and no luck.

How do I properly return this socket file?

2 Answers



At first glance, the above is the key indicator. Sockets aren't accessed over http locally, and that line seems to state that it's attempting to do so -- when it does, it fails.

Socket connects should look like:


So you'd need to find the line where the http:// is being defined and remove that.

  • Thanks for the feedback I do not believe that is.

    That comes right out of the Digital Ocean tutorial (linked in post). As part of /etc/nginx/sites-available/myproject under server { listen {} } it has proxy_pass http://unix:/home/user/myproj/proj.sock;

    That's what I had before and it worked. That's what I have on a similar droplet with another flask application that works.

    But to be sure I did replace it to just proxy_pass unix:/home/sammy/myproject/myproject.sock; and that throws an error. I cannot restart nginx as I receive the error nginx: [emerg] invalid URL prefix

    • @rbourne

      Ok, what I did was spin up a Droplet and ran through the entire guide twice to make sure I was able to duplicate what I did.

      The one fail point is with permissions on the /home/sammy directory. I had a general idea that this would be where I'd land, but I wanted to make sure since I am not a Python dev by trade (I normally work with PHP and Bash), thus my working knowledge of the setup other than NGINX is a bit limited.

      What you should need to do is run:

      chown -R sammy:www-data /home/sammy

      and then restart myproject and nginx ...

      sudo systemctl restart myproject \
      && sudo systemctl restart nginx

I found the solution.

My uploaded app.py was not able to be compiled on the server due to having not (yet) installed the requirements. I used pip install from my requirements.txt, verified I could run python app.py and when it was good to go I restarted myproject.service and the socket file was created. I could then view the app on the internet.

Have another answer? Share your knowledge.