nut_webgui binary seem unable to daemonize on linux Alpine/musl

[integratec]

First of all, thank you for this simple and efficient software!
I installed it on Alpine Linux along with NUT (nethserver role).
I did everything manually because the provided install script doesn't support OpenRC and also fail because doesn't even support Busybox's sha256 verification.
I prepared a small OpenRC-compatible basic script to run at system startup.

#!/sbin/openrc-run

name="NUT Webgui"
description="nut_webgui - Simple NUT Web interface"

depend() {
    use net
}

command="/usr/local/bin/nut_webgui"
command_args="--config-file /etc/nut_webgui/config.toml --allow-env"
required_files="/etc/nut_webgui/config.toml /etc/nut_webgui/users.toml"

# The special RC_SVCNAME variable contains the name of this service.
pidfile="/run/${RC_SVCNAME}.pid"
# Seems does not know how to background iself here
command_background=true

If I omit the last directive in the above script, the application starts and runs correctly, but it doesn't go to background.
Am I doing something terribly wrong, or is something preventing it from working properly when using musl as a system library? Thanks.

[SuperioOne]

Thanks!

install.sh script indeed requires some additional care. I need to strip out some GNU specific extensions and add more init system support.

I’ve just created an Alpine VM (version 3.24) and installed nut_webgui v0.10.2 x86_64 musl binary.

My init file is almost exact the same. I only changed depend to need net, but use net also works for me.

/etc/init.d/nut_webgui

#!/sbin/openrc-run

name="NUT Webgui"
description="nut_webgui - SImple NUT Web Interface"

depend() {
    need net
}

command="/usr/local/bin/nut_webgui"
command_args="--config-file /etc/nut_webgui/config.toml --allow-env"
required_files="/etc/nut_webgui/config.toml /etc/nut_webgui/users.toml"
pidfile="/run/${RC_SVCNAME}.pid"
command_background=true

and I added it to the default runlevel

rc-update add nut_webgui

So far, I can access the web page from another PC. Of course, my test uses the root user as the service owner.

I’m not very familiar with OpenRC, but if your service config uses a different user (rc-update --user), it could be related to a missing lingering mechanism for that user. Another thing is config file permission issues, although that also wouldn’t work with command_background=false.

[integratec]

Thanks.
Anyway, the main question was whether it's necessary to use command_background=true or not. Maybe I expressed myself poorly; sorry, but English isn't my native language.
The OpenRC docs say:

1. If the daemon backgrounds itself and creates its own PID file, use pidfile.
2. If the daemon does not background itself (or has an option to run in the foreground) and does not create a PID file, then use command_background=true and pidfile.
3. ...

I expected the nut_webgui daemon both backgrounds itself and does create a pidfile. So I thought this wasn't necessary but I was wrong because the firs option was not enough but the secon one worked.

About use net/need net: the latter is not satisfied by the loopback interface so I used the first one to relax dependency.

[SuperioOne]

Yes, command_background=true is necessary. nut_webgui does not daemonize/background itself.