Could not access seelf with env credentials

[decryptedchaos]

After six hours of tinkering i can only conclude this is woefully broken? Firstly even just exposed directly via port 8080 the thing doesn't work it wont accept the login defined in the envar HTTP 400 error

Beyond that, it seems the traefik integration is also broken? i ended up rolling my own traefik config to work around it, but your claim of painless is lacking..

I've had enough of trying to make it work for now, ask me for what would be helpful and i'll try to provide it tomorrow.

[YuukanOO]

Could you please provide more informations such as the installation method?

Firstly even just exposed directly via port 8080 the thing doesn't work it wont accept the login defined in the envar

The account details provided in the environment is only used to create the first account. Once the first account has been set up, if you want to change the admin password, you must use the profile page.

traefik integration is also broken

The traefik part is deployed only when creating a target or when exposing seelf using a local target.

[decryptedchaos]

Could you please provide more informations such as the installation method?

It is deployed via docker compose using the following (slightly modified for my own traefik workaround

services:
  web:
    restart: unless-stopped
    image: yuukanoo/seelf:2.4.0 
    container_name: seelf
    environment:
      - ADMIN_EMAIL=redacted-email # Its set correctly 
      - ADMIN_PASSWORD=redacted-password # as shown here
      - EXPOSED_ON=http://seelf@mydomain.tld # <- I've done it this way
      - EXPOSED_ON=http://seelf@docker.localhost # <- I've done it this way as a test
      - HTTP_SECURE= # Force fallback to the default handling of http secure (based on EXPOSED_ON if set)
   # ports:
   #   - "8080:8080"
    labels:
      - app.seelf.exposed=true
      - app.seelf.subdomain=seelf
      - subdomain=seelf # this is my workaround for a standalone traefik
      - domain=mydomain.tld # also my work around
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - ssh:/root/.ssh 
      - data:/seelf/data 
    networks:
      - proxy

networks:
  proxy:
    external: true
volumes:
  ssh:
  data:

The account details provided in the environment is only used to create the first account. Once the first account has been set up, if you want to change the admin password, you must use the profile page.

Right, but since it wont accept the one i supply in the compose file i can never login in the first place.

This is what i get in the browser console:

The traefik part is deployed only when creating a target or when exposing seelf using a local target.

Yes as you can see i've done this and it does NOT work, it never exposes itself..

[YuukanOO]

Could you start from scratch by doing a docker compose down --volumes to make sure volumes are completely removed and then trying with something along the lines:

services:
  web:
    restart: unless-stopped
    image: yuukanoo/seelf:2.4.0
    container_name: seelf
    environment:
      - ADMIN_EMAIL=your-email
      - ADMIN_PASSWORD=your-password
      - EXPOSED_ON=http://seelf@your-domain.tld
      - HTTP_SECURE=
    labels: 
      - app.seelf.exposed=true
      - app.seelf.subdomain=seelf
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - ssh:/root/.ssh
      - data:/seelf/data

volumes:
  ssh:
  data:

[YuukanOO]

Another idea that comes to my mind is if the target setup fail (when deploying traefik), it will fail silently in this particular case, I should probably just returns the error here.

If my above solution is not working, maybe you can try with only ports binding:

services:
  web:
    restart: unless-stopped
    image: yuukanoo/seelf:2.4.0
    environment:
      - ADMIN_EMAIL=your-email
      - ADMIN_PASSWORD=your-password
    ports:
      - "8080:8080"
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - ssh:/root/.ssh
      - data:/seelf/data

volumes:
  ssh:
  data:

And check if you can access it on http://your-domain.tld:8080. Then create a local target and check if it succeeds.

[decryptedchaos]

Could you start from scratch by doing a docker compose down --volumes to make sure volumes are completely removed and then trying with something along the lines:

Okay, so this did in fact work, so like you say the issue here is no feedback is given, it just starts up and never says anything about their being a issue with the deployment, and there was clearly another issue with the login credentials in the volume that was overriding what was supplied in the compose file.

This in my view is quite a problem, it creates a admittedly rare situation like i had where there was some error deploying the first account and after that first attempt it never attempted to re-deploy the account i don't know how off the top of my head but there really should be some kind of handler for this case.

Either by a env value for RESET_LOGIN=yes or something like that because its not a standard docker workflow, its typically assumed that using envars are volatile and will get reset on every launch, since you appear to be using them only on the first run, its not whats typically assumed about docker deployments.

[YuukanOO]

Glad it worked! I will push a fix in few minutes to return the error when it fails to deploy the target configured through EXPOSED_ON to make sure it will not insert the target if it can't make sure it has been correctly configured.

As for the account, I'm not sure what happened :/ Did you start seelf a first time and then changed the password in the compose file without removing volumes?

since you appear to be using them only on the first run

Indeed, they are used only if no account already exists to avoid having to deal with a setup page.

[decryptedchaos]

Glad it worked! I will push a fix in few minutes to return the error when it fails to deploy the target configured through EXPOSED_ON to make sure it will not insert the target if it can't make sure it has been correctly configured.

Sounds good!

As for the account, I'm not sure what happened :/ Did you start seelf a first time and then changed the password in the compose file without removing volumes?

AFAIK no, and if i had it would have been the default values from the docs admin@example.com / admin and i tried using those just to make sure i didn't do that. it rejected that too and i wouldn't have put anything else in there.

P.S I STRONGLY suggest this gets put into the docs somewhere in a infobox that says something along the lines of if your first account fails to deploy on the first run delete the volumes and retry had the docs said that i probably could have figured it out, but there really wasn't anything to definitively imply that the credentials were stored in the volume and it should be deleted

[YuukanOO]

admin@example.com / admin and i tried using those just to make sure i didn't do that. it rejected that too and i wouldn't have put anything else in there.

That was my next question :)

Either way, the fix is on its way so thanks for reporting it.

I'll leave the issue open for now to remember to think about what could have gone wrong.