[BUG] PUID / PGID does not work with nginx

[aleclerc7]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

Hello,

When I set a PUID / PGID (for use with a volume),

• all the file ownership are set to that PUID / PGID.
• Very good. 👍

But the nginx server still wants to run with user abc (911) and is not able to read the files.

• HTTP 500 error. 👎

If I set only the PGID:

• Ownership becomes "911" on the host system.
• Then nginx works, and because of the group 100, other users of the host system can access the files.

But this is very problematic.

• How can nginx be made to work properly and have read/write access when a PUID is set?

I suspect this is a bug; something not set properly in the docker image.

Thank you very much for your kind help!

Expected Behavior

The nginx server should be able to read the files to serve the wbeside even if the PUID PGID are set to something.

Right now this is a : HTTP ERROR 500

Steps To Reproduce

I take the docker compose proposed by default (in the doc), and set the PUID and GUID, and set the volume path.

---
services:
  grav:
    image: lscr.io/linuxserver/grav:latest
    container_name: Grav-MyWebSite
    environment:
      - PUID=1038
      - PGID=100
      - TZ=America/New_York
    volumes:
      - /volume1/my_web_site:/config
    ports:
      - 47801:80
    restart: unless-stopped

Environment

Synology NAS DMS 7.2.2
Portainer 2.21.4, using stacks for the docker compose.

I do not know which version of the docker image I use. But it is recent. As far as I know the nginx and site conf are the default ones. I attached the files.

env:

root@c838b280e5dd:/# env
PUID=1038
HOSTNAME=c838b280e5dd
S6_CMD_WAIT_FOR_SERVICES_MAXTIME=0
PWD=/
TZ=America/New_York
HOME=/root
LANG=C.UTF-8
PGID=100
VIRTUAL_ENV=/lsiopy
S6_VERBOSITY=1
S6_STAGE2_HOOK=/docker-mods
TERM=xterm-256color
SHLVL=1
PS1=$(whoami)@$(hostname):$(pwd)\$ 
LC_ALL=C.UTF-8
LSIO_FIRST_PARTY=true
PATH=/lsiopy/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
_=/usr/bin/env

nginx.conf.txt

default.conf.txt

CPU architecture

x86-64

Docker creation

Container logs

[migrations] started
[migrations] 01-nginx-site-confs-default: skipped
[migrations] 02-default-location: skipped
[migrations] done
───────────────────────────────────────
      ██╗     ███████╗██╗ ██████╗
      ██║     ██╔════╝██║██╔═══██╗
      ██║     ███████╗██║██║   ██║
      ██║     ╚════██║██║██║   ██║
      ███████╗███████║██║╚██████╔╝
      ╚══════╝╚══════╝╚═╝ ╚═════╝
   Brought to you by linuxserver.io
───────────────────────────────────────
To support the app dev(s) visit:
Grav: https://opencollective.com/grav/donate
To support LSIO projects visit:
https://www.linuxserver.io/donate/
───────────────────────────────────────
GID/UID
───────────────────────────────────────
User UID:    1038
User GID:    100
───────────────────────────────────────
using keys found in /config/keys
[custom-init] No custom files found, skipping...
[ls.io-init] done.

[github-actions]

A human has marked this issue as invalid, this likely happened because the issue template was not used in the creation of the issue.

[aleclerc7]

So where can I go for help if this is not a problem with the built-in nginx server?

[j0nnymoe]

It's marked as invalid because you haven't provided any details for us to replicate the error and zero logs.

[aleclerc7]

I updated the bug report with more information. If this is not enought, tell me.

[aptalca]

If I set only the PGID:
Ownership becomes "911" on the host system.

That is expected as 911 is the default value used for PUID/PGID if the user does not set them

But the nginx server still wants to run with user abc (911) and is not able to read the files.

Not clear what you mean.

Main nginx service runs as root. The workers run as user abc, which is set to the PUID/PGID used. The workers will only run as 911 if you don't set PUID

[aleclerc7]

Hello @aptalca,

Thank you for your answer.

Yeah, this sentence is not clear. What I mean is that when I set the PUID, in the volume (actual NAS files), the ownership of the files is correctly set to that PUID/PGID. (If not set, it is set to 911.) This is all fine.

But for a reason I do not know, HTTP ERROR 500 is the only result afterward. All files have the good owner/group. But nginx is unable to read the files (this is my analysis, which could be wrong). But it looks like an access issue.

When I comment out the PUID and only keep the PGID, then the ownership of all the file as 911/PGID. Fine, again.

And then nginx is able to serve the GravCMS website.

So I think there is a setup issue in the docker??? When PUID is set, nginx (and its workers) should use this PUID to be able to read the files since all the files are now owned by this user/group. Or at least be in the PGID group???

Thank you for you help! 😊

[aptalca]

Ah, I think*know what the issue is. We recently fixed a bug in the baseimage where we got rid of unnecessary chowns. But that has a negative effect on this container because it's not fixing the perms on the nginx files in the rest of the config folder: https://github.com/linuxserver/docker-grav/blob/main/root/etc/s6-overlay/s6-rc.d/init-grav-config/run#L49-L51

It only fixes the www folder.

We'll get this fixed. In the meantime, you can recursively chown the whole config folder for the PUID you want.

Thanks for letting us know.

[aleclerc7]

Hello @aptalca,

I might have an older image then. I confirm that the owner is everywhere.

But I found my beginner problem: when I shared the folder I did give my user (UID) the permission to access it. But I just saw that even if all the files were owned by the UID, there were no permission set at all for the files for the user (UID). I explicitly set the permission, and now all is working fine.

Thank you very much for your time and help. The error was on my side (as often the case with users!).

I cannot tell if there is something more but for the docker I run, setting the actual permission made it (without changing the owner, as it looked fine everywhere).

Best regards!