Elastic Beanstalk部署Django Channels报错：虚拟环境路径不存在

Problem Description #

I'm following a tutorial to deploy a Django Channels app on AWS Elastic Beanstalk, using the following configuration files:

01_env.config #

option_settings:
  aws:elasticbeanstalk:container:python:
    WSGIPath: "dashboard/wsgi.py"
  aws:elasticbeanstalk:application:environment:
    DJANGO_SETTINGS_MODULE: "dashboard.settings"
    PYTHONPATH: /opt/python/current/app/dashboard:$PYTHONPATH
  aws:elbv2:listener:80:
    DefaultProcess: http
    ListenerEnabled: 'true'
    Protocol: HTTP
    Rules: ws
  aws:elbv2:listenerrule:ws:
    PathPatterns: /websockets/*
    Process: websocket
    Priority: 1
  aws:elasticbeanstalk:environment:process:http:
    Port: '80'
    Protocol: HTTP
  aws:elasticbeanstalk:environment:process:websocket:
    Port: '5000'
    Protocol: HTTP

02_setup.config #

container_commands:
  00_pip_upgrade:
    command: "source /opt/python/run/venv/bin/activate && pip install --upgrade pip"
    ignoreErrors: false
  01_migrate:
    command: "django-admin.py migrate"
    leader_only: true
  02_collectstatic:
    command: "django-admin.py collectstatic --noinput"
  03_wsgipass:
    command: 'echo "WSGIPassAuthorization On" >> ../wsgi.conf'

When running eb create django-env, the deployment fails with the error:

Command failed on instance. An unexpected error has occurred [ErrorCode: 0000000001]

Checking the logs, I found the root cause:

2020-06-17 16:36:41,880 P4189 [INFO] Command 00_pip_upgrade
2020-06-17 16:36:41,883 P4189 [INFO] -----------------------Command Output-----------------------
2020-06-17 16:36:41,883 P4189 [INFO] /bin/sh: /opt/python/run/venv/bin/activate: No such file or directory

I can't SSH into the EC2 instance to investigate further. Has the Python virtual environment directory changed in Amazon Linux 2?

Solution #

Yes, the virtual environment path has changed significantly between the legacy Amazon Linux AMI and Amazon Linux 2 on Elastic Beanstalk. The tutorial you followed was written for the old AMI, which uses the hardcoded path /opt/python/run/venv/bin/activate — this path does not exist on Amazon Linux 2.

Key Changes to Fix the Issue #

1. Remove manual virtual environment activation
   Elastic Beanstalk's container_commands for Python on Amazon Linux 2 automatically run within the application's virtual environment. You don't need to explicitly source an activate script.

2. Update your 02_setup.config file
   Replace your existing container commands with this revised version, which is compatible with Amazon Linux 2:

   container_commands:
     00_pip_upgrade:
       command: "pip install --upgrade pip"
       ignoreErrors: false
     01_migrate:
       command: "python manage.py migrate"
       leader_only: true
       cwd: "/opt/python/current/app"
     02_collectstatic:
       command: "python manage.py collectstatic --noinput"
       cwd: "/opt/python/current/app"
     03_wsgipass:
       command: 'echo "WSGIPassAuthorization On" >> ../wsgi.conf'
   

Additional Notes #

• Why this works:

  • The pip and python commands in container_commands point directly to the virtual environment's binaries (no activation needed).
  • Adding cwd: "/opt/python/current/app" ensures commands run from your application's root directory, where manage.py resides.
  • If you ever need to manually activate the venv for custom scripts (rare), use a dynamic path instead of hardcoding:

    source $(find /var/app/venv -name activate)
    

    Amazon Linux 2 stores virtual environments under /var/app/venv/ with a randomly generated suffix (e.g., /var/app/venv/staging-LQM1lest/bin/activate), so hardcoding the path will always fail.

• Verify other configuration:
  Your 01_env.config looks correct for routing WebSocket traffic, but ensure your Django Channels setup (e.g., ASGI_APPLICATION, channel layer configuration) is compatible with Elastic Beanstalk's process mapping for port 5000.

内容的提问来源于stack exchange，提问作者Filipe Aleixo