Railway•3mo ago

Retrieving Django database sessions work locally behind a proxy but fail on Railway?

Does anyone know if or why reading Django (DRF) sessions might fail behind a reverse proxy on Railway such as Caddy or Nginx but work locally? I am using django.contrib.sessions.backends.db in my case.

37 Replies

Percy•3mo ago

Project ID: N/A

ChrisHuppert•3mo ago

N/A

Brody•3mo ago

are you having gunicorn trust the proxy headers?

ChrisHuppert•3mo ago

I will be honest, I don't think I have done that. Do you know where I can find documentation for this?

Brody•3mo ago

https://docs.gunicorn.org/en/stable/settings.html?highlight=forwarded-allow-ips#forwarded-allow-ips add the flag --forwarded-allow-ips * to your gunicorn start command

ChrisHuppert•3mo ago

Thanks! I will try that out and see if it works So when I run this I get an error:

Failed to find attribute 'application' in 'Core'.

Failed to find attribute 'application' in 'Core'.

I seen online that it may have something to do with the wsgi.py file but I am not quite sure. Here is my wsgi.py file:

"""
WSGI config for django_project project.

It exposes the WSGI callable as a module-level variable named ``application``.

For more information on this file, see
https://docs.djangoproject.com/en/4.2/howto/deployment/wsgi/
"""

import os
from dotenv import load_dotenv, find_dotenv

load_dotenv(dotenv_path=find_dotenv(filename='.env'))

from django.core.wsgi import get_wsgi_application

os.environ.setdefault('DJANGO_SETTINGS_MODULE', f'settings.{os.getenv("DJANGO_SETTINGS_FILE")}')

application = get_wsgi_application()

"""
WSGI config for django_project project.

It exposes the WSGI callable as a module-level variable named ``application``.

For more information on this file, see
https://docs.djangoproject.com/en/4.2/howto/deployment/wsgi/
"""

import os
from dotenv import load_dotenv, find_dotenv

load_dotenv(dotenv_path=find_dotenv(filename='.env'))

from django.core.wsgi import get_wsgi_application

os.environ.setdefault('DJANGO_SETTINGS_MODULE', f'settings.{os.getenv("DJANGO_SETTINGS_FILE")}')

application = get_wsgi_application()

Brody•3mo ago

what's your start command

ChrisHuppert•3mo ago

exec gunicorn --forwarded-allow-ips * django_project.wsgi:application --bind [::]:${BACKEND_DJANGO_PORT} --timeout 0

exec gunicorn --forwarded-allow-ips * django_project.wsgi:application --bind [::]:${BACKEND_DJANGO_PORT} --timeout 0

Brody•3mo ago

and where are you defining that start command

ChrisHuppert•3mo ago

Oh that is in a set_up.sh file that is run by a Dockerfile

Brody•3mo ago

do you need to talk to your django app over the private network?

ChrisHuppert•3mo ago

That will be the endgame plan

Brody•3mo ago

use this then -

exec gunicorn django_project.wsgi --bind [::]:${PORT} --timeout 0 --forwarded-allow-ips *

exec gunicorn django_project.wsgi --bind [::]:${PORT} --timeout 0 --forwarded-allow-ips *

side note, always best to use the standard names for environment variables such as PORT, then since you will need a fixed port simply define PORT as a service variable

ChrisHuppert•3mo ago

Ahh thanks for the information. I will reploy now Unfortunately I get the same error:

Failed to find attribute 'application' in 'Core'.

Failed to find attribute 'application' in 'Core'.

I must be doing something wrong:

gunicorn --forwarded-allow-ips * django_project.wsgi --bind [::]:${PORT} --timeout 0

gunicorn --forwarded-allow-ips * django_project.wsgi --bind [::]:${PORT} --timeout 0

Brody•3mo ago

do you have the start command set elsewhere, like in the service settings?

ChrisHuppert•3mo ago

Sorry, where would I find the service settings? I am not using a procfile or a gunicorn config file

Brody•3mo ago

you can find the settings tab after opening your service

ChrisHuppert•3mo ago

Oh you mean in railway Actually it fails locally also And there is no additional start command in my railway settings

Brody•3mo ago

share your dockerfile?

ChrisHuppert•3mo ago

FROM python:3.11.6

WORKDIR /app/django_project

COPY ./backend-django/requirements.txt ./

RUN pip install --no-cache-dir -r requirements.txt

COPY ./backend-django /app

WORKDIR /app/django_project

EXPOSE 8000
EXPOSE 5672
EXPOSE 15672

# RUN python manage.py makemigrations && python manage.py migrate

CMD ["sh", "set_up.sh"]

FROM python:3.11.6

WORKDIR /app/django_project

COPY ./backend-django/requirements.txt ./

RUN pip install --no-cache-dir -r requirements.txt

COPY ./backend-django /app

WORKDIR /app/django_project

EXPOSE 8000
EXPOSE 5672
EXPOSE 15672

# RUN python manage.py makemigrations && python manage.py migrate

CMD ["sh", "set_up.sh"]

And set_up.sh:

exec gunicorn --forwarded-allow-ips * django_project.wsgi --bind [::]:${PORT} --timeout 0

exec gunicorn --forwarded-allow-ips * django_project.wsgi --bind [::]:${PORT} --timeout 0

Brody•3mo ago

no railway.json either?

ChrisHuppert•3mo ago

No sir

Brody•3mo ago

is django_project the correct folder name?

ChrisHuppert•3mo ago

Maybe it is how my project structure is structured?

Brody•3mo ago

share your repo?

ChrisHuppert•3mo ago

Here is my repository structure:

backend-django
  django_project
    AppTwo
      ...
    Core
      API
        v1
          __init__.py
          admin.py
          apps.py
          helpers.py
          permisssions.py
          serializers.py
          services.py
          views.py
          viewsets.py
        v2
          (similar to v1)
      migrations
      ...
    django_project
      __init__.py
      asgi.py
      router.py
      urls.py
      wsgi.py
    settings
      __init__.py
      base.py
      settings_local.py
      settings_production.py
    set_up.sh

backend-django
  django_project
    AppTwo
      ...
    Core
      API
        v1
          __init__.py
          admin.py
          apps.py
          helpers.py
          permisssions.py
          serializers.py
          services.py
          views.py
          viewsets.py
        v2
          (similar to v1)
      migrations
      ...
    django_project
      __init__.py
      asgi.py
      router.py
      urls.py
      wsgi.py
    settings
      __init__.py
      base.py
      settings_local.py
      settings_production.py
    set_up.sh

Brody•3mo ago

quite hard to get a good grasp of what's going on like that, if sharing a link to your repo not a option, could you add me to it?

ChrisHuppert•3mo ago

Unfortunately I had signed an NDA so it is not possible at the moment. But if you would like to look at a specific file I can do my best to provide it

Brody•3mo ago

ah okay no worries I'm kinda out of ideas tbh you said running the same set_up.sh locally gives you the same error?

ChrisHuppert•3mo ago

That is correct

Brody•3mo ago

are you running that script while in the same folder that the script resides in?

ChrisHuppert•3mo ago

This is also correct

Brody•3mo ago

back to having no ideas

ChrisHuppert•3mo ago

Ah okay I fixed it. It was actually a syntax error where I did not put "*" in:

exec gunicorn --forwarded-allow-ips * django_project.wsgi --bind [::]:${PORT} --timeout 0

exec gunicorn --forwarded-allow-ips * django_project.wsgi --bind [::]:${PORT} --timeout 0

Basically it was a string

Brody•3mo ago

what a fun little bug

Brody•3mo ago

it did give an example, yet i gave you the wrong start command, my bad

ChrisHuppert•3mo ago

No problem! I am going to figure out why the Django application is not using the django_session table

Gaming

Programming

Retrieving Django database sessions work locally behind a proxy but fail on Railway?