Fresh Docker Install, loads UI but won't load Available Sources

[alejakun]

Describe the bug

I just installed with the docker image, it seems to work fine, however it does not show anything when trying to add a New Library/Manga in the field Available sources. Checking at the logs it shows the following:

{"level":50,"time":1696973407106,"pid":308,"hostname":"d12d3bc2e27a","msg":"Failed to get available sources. err: Error: Command failed with exit code 2: mangal sources list -r\npanic: mkdir /config/.config/mangal/sources: permission denied\n\ngoroutine 1 [running]:\ngithub.com/samber/lo.must({0xefeae0, 0xc000476d80}, {0x0, 0x0, 0x0})\n\tgithub.com/samber/lo@v1.37.0/errors.go:53 +0x1f4\ngithub.com/samber/lo.Must0(...)\n\tgithub.com/samber/lo@v1.37.0/errors.go:72\ngithub.com/metafates/mangal/where.mkdir({0xc000265aa0, 0x1e})\n\tgithub.com/metafates/mangal/where/where.go:18 +0x57\ngithub.com/metafates/mangal/where.Sources()\n\tgithub.com/metafates/mangal/where/where.go:39 +0x5b\ngithub.com/metafates/mangal/provider.Customs()\n\tgithub.com/metafates/mangal/provider/provider.go:31 +0x49\ngithub.com/metafates/mangal/cmd.glob..func24.3()\n\tgithub.com/metafates/mangal/cmd/sources.go:67 +0x3b\ngithub.com/metafates/mangal/cmd.glob..func24(0x1d291e0, {0x101d881?, 0x1?, 0x1?})\n\tgithub.com/metafates/mangal/cmd/sources.go:82 +0x3ea\ngithub.com/spf13/cobra.(*Command).execute(0x1d291e0, {0xc0004a0640, 0x1, 0x1})\n\tgithub.com/spf13/cobra@v1.6.1/command.go:920 +0x847\ngithub.com/spf13/cobra.(*Command).ExecuteC(0x1d2a040)\n\tgithub.com/spf13/cobra@v1.6.1/command.go:1044 +0x3bd\ngithub.com/spf13/cobra.(*Command).Execute(...)\n\tgithub.com/spf13/cobra@v1.6.1/command.go:968\ngithub.com/metafates/mangal/cmd.Execute()\n\tgithub.com/metafates/mangal/cmd/root.go:112 +0x7a\nmain.main()\n\tgithub.com/metafates/mangal/main.go:13 +0x4d\nMangadex\nManganelo\nManganato\nMangapill"}

Reproduction steps

Fresh install with the following docker compose:

  kaizoku:
    image: ghcr.io/oae/kaizoku:latest
    container_name: kaizoku
    restart: unless-stopped
    security_opt:
      - no-new-privileges:true
    networks:
      - traefik
    ports:
      - "$KAIZOKU_PORT:3000"
    environment:
      - DATABASE_URL=postgresql://kaizoku:kaizoku@kaizoku_db:5432/kaizoku
      - KAIZOKU_PORT=3000
      - REDIS_HOST=kaizoku_redis
      - REDIS_PORT=6379
      - PUID=$PUID
      - PGID=$PGID
      - TZ=$TZ
    volumes:
      - $PWD/kaizoku/config:/config
      - $PWD/kaizoku/data:/data
      - $PWD/kaizoku/logs:/logs
    depends_on:
      kaizoku_db:
        condition: service_healthy

  kaizoku_redis:
    image: redis:7-alpine
    container_name: kaizoku_redis
    user: "$PUID:$PGID"
    restart: unless-stopped
    security_opt:
      - no-new-privileges:true
    networks:
      - traefik
    volumes:
      - $PWD/kaizoku/redis:/data
  kaizoku_db:
    image: postgres:15-alpine
    container_name: kaizoku_db
    user: "$PUID:$PGID"
    restart: unless-stopped
    security_opt:
      - no-new-privileges:true
    networks:
      - traefik
    healthcheck:
      test: ['CMD-SHELL', 'pg_isready -U kaizoku']
      interval: 5s
      timeout: 5s
      retries: 5
    environment:
      - POSTGRES_USER=kaizoku
      - POSTGRES_DB=kaizoku
      - POSTGRES_PASSWORD=kaizoku
    volumes:
      - $PWD/kaizoku/postgres:/var/lib/postgresql/data

Expected behavior

Should load something for Available sources.

Additional context

No response

[notHiks]

you have a permisison error. the docker u8ser cazn't write to the filesystem

mkdir /config/.config/mangal/sources: permission denied\n\ngoroutine 1

[alejakun]

Hi, thanks for the reply. I totally agree, however I see no reason why this is happening, since the user was able to create the whole filesystem structure starting from config (ie. /config/.config/mangal/sources), with what is in fact the user I specified in the compose environment. Just to be sure I deleted the directories and started again (easy since docker), but I still have the same behavior:

The docker user creates the directories structure with correct user/group information but sends the same error.

One thing I noticed which strikes me as odd (not necessarily wrong, just odd), is the permissions for file: /config/.config/mangal/mangal.toml which was created (again fresh run), with user ownership as: root:root

Any thoughts would be greatly appreciated.

[notHiks]

root should not own that file. for my install all files in /config are owned by the same user.

[alejakun]

Are you using the Docker version?

[notHiks]

yes i'm running it on docker. i'm also pointing the files to a NAS so all ownership is tied to the same user id's.

[alejakun]

It should be the same for me, or at least it’s what I’m expecting with the environment variables set to ($PUID and $PGID have correct values):

      - PUID=$PUID
      - PGID=$PGID

[alejakun]

@notHiks , would you mind sharing your docker-compose or docker command just to compare?

[alejakun]

At the end it seems to have problemas when you set the user:group information (at least in my case). In the end I just commented the lines and restarted docker and it worked, however I would not consider this as closed since the docker compose explicitly stated the options and I would not work like that (which would be better since I'm not really a fan of letting a container run as root).

Best regards.

[notHiks]

a run it behind a vpn and the storage is on a NAS owned by user 1000

services:
  app:
    container_name: kaizoku
    network_mode: "container:gluetun"
    image: ghcr.io/oae/kaizoku:latest
    environment:
      - DATABASE_URL=postgresql://kaizoku:kaizoku@0.0.0.0:5432/kaizoku
      - KAIZOKU_PORT=3000
      - REDIS_HOST=0.0.0.0
      - REDIS_PORT=6379
      - PUID=1000
      - PGID=1000
      - TZ=Europe/London
    volumes:
      - /home/appdata/storage/Manga:/data
      - /home/appdata/storage/.kaizoku/config:/config
      - /home/appdata/storage/.kaizoku/logs:/logs
    depends_on:
      db:
        condition: service_healthy
    security_opt:
      - no-new-privileges:true
   # ports:
   #   - '3000:3000'

  redis:
    container_name: kaizoku-redis
    image: redis:7-alpine
    network_mode: "container:gluetun"
    restart: unless-stopped
    volumes:
      - redis:/data
    security_opt:
      - no-new-privileges:true

  db:
    container_name: kaizoku-db
    image: postgres:alpine
    network_mode: "container:gluetun"
    restart: unless-stopped
    security_opt:
      - no-new-privileges:true
    healthcheck:
      test: ['CMD-SHELL', 'pg_isready -U kaizoku']
      interval: 5s
      timeout: 5s
      retries: 5
    environment:
      - POSTGRES_USER=kaizoku
      - POSTGRES_DB=kaizoku
      - POSTGRES_PASSWORD=kaizoku
    volumes:
      - db:/var/lib/postgresql/dat

[alejakun]

I swear I double checked my permissions, but with all the containers maybe they were al messed up, it's working now with my user, sorry for the time.

Best Regards.