BUG: Restarting docker logs users out

[adamchemiawppl]

Hi,

Not sure it is "expected", or not, but every time I restart/rebuild container, all users get logged out.
I'm running Wordpress with Mariadb and Redis (non-persistent).

Any idea what is the root cause? Is there a chance Redis actually handles login tokens (then restarting it would obviously remove such data, if it is not saved in DB). Then again - it seems with redis plugin disabled the same thing happens.

Any idea what to look for? Couldn't find any ideas on the root cause.

Best regards,
Adam Wojciechowski

[tianon]

This is something I've noticed as well, but I never quite figured out exactly what's causing it.

I think it's related to PHP sessions and where they get stored (might need to set session.save_path to a persistent location?).

[adamchemiawppl]

Well, AFAIK wordpress manages sessions by itself, rather than using php in-built sessions, so I think it shouldn't be related.

[adamchemiawppl]

Any chance for additional input from anybody? It is a fairly significant issue, impacting usefulness of this repo.

[LaurentGoderre]

I can't reproduce with just Wordpress and MariaDB

[LaurentGoderre]

Still not able to reproduce with

version: '3.1'

services:

  wordpress:
    image: wordpress
    restart: always
    ports:
      - 8080:80
    environment:
      WORDPRESS_DB_HOST: db
      WORDPRESS_DB_USER: exampleuser
      WORDPRESS_DB_PASSWORD: examplepass
      WORDPRESS_DB_NAME: exampledb
      WORDPRESS_CONFIG_EXTRA: define('WP_REDIS_HOST', 'redis');
    volumes:
      - wordpress:/var/www/html

  db:
    image: mariadb
    restart: always
    environment:
      MARIADB_DATABASE: exampledb
      MARIADB_USER: exampleuser
      MARIADB_PASSWORD: examplepass
      MARIADB_ROOT_PASSWORD: '1'
    volumes:
      - db:/var/lib/mysql

  redis:
      image: 'redis:alpine'
      ports:
        - '6379:6379'
      restart: always
      expose:
        - '6379'

volumes:
  wordpress:
  db:

[adamchemiawppl]

Interesting. On most recent deployment it also worked correctly on my end, so it seems it was somehow fixed. Before that it logged all users out on over 50 deployments during development.

Please don't close it yet - in next 2 days I'll have additional deployments & will update the ticket if solved/unsolved.

[adamchemiawppl]

Hi,

Sorry, it turned out to be a dumb mistake. WP salts were not set properly - therefore they were randomly generated whenever container was restarted. Thanks for support, sorry for your time - and hopefully if anybody encounters such issue, my case will help them solve it :-)