Hawser service starts in a crashed state when host is rebooted

[xenogenick]

I am running Hawser in a fully up-to-date Alpine v3.32.3 LXC container on Proxmox that hosts a simple docker application, and it works great, does exactly what I want it to for using Dockhand across my environment.

The problem is that when I reboot the Alpine host (I have this on any I've attempted, and have given this container up to 1GB of memory and 2 cores to no avail), the Hawser service starts in a 'crashed' state and I have to manually restart it each time, which is not ideal. The service is enabled and in the 'default' runlevel.

My Hawser config file just contains DOCKER_SOCKET=/var/run/docker.sock and PORT=2376 settings (no token for now), and the service file is the default as per the installation script:

#!/sbin/openrc-run
name="hawser"
description="Hawser - Remote Docker Agent for Dockhand"
command="/usr/local/bin/hawser-wrapper"
command_background="yes"
pidfile="/run/${RC_SVCNAME}.pid"
start_stop_daemon_args="--stdout /var/log/hawser.log --stderr /var/log/hawser.log"
depend() {
    need net docker
    after docker
}

The log file says Error loading config: Docker socket not found at /var/run/docker.sock (sometimes more than once) even though the service file is depending on docker and is supposed to be starting after it.

In my /etc/rc.conf I have set rc_depend_strict=YES but it makes no difference.

I am fully expecting for this to be something stupid that I've done somewhere, but I've been trying to fix it for a while with no success so here I am!

[riskwise99]

You have same problem I'm having and I'm starting to give up mate! The latest version isn't helping either.
OpenRC - you are right to put in a delay on startup, let docker start up wait a bit then bring up hawser.
But I've found without touching docker, after a day or so, I can't manage the docker instance, because hawser has simply just crashed out.

Here's my init.d ..

#!/sbin/openrc-run

name="hawser"
description="Hawser - Remote Docker Agent for Dockhand"
command="/usr/local/bin/hawser-wrapper"
command_background="yes"
pidfile="/run/${RC_SVCNAME}.pid"


command_args_background="--stdout /var/log/${RC_SVCNAME}.log --stderr /var/log/${RC_SVCNAME}.log"

depend() {
    need localmount net docker
    after docker
}

start_pre() {
    # Ensure the log file exists so start-stop-daemon can write to it
    checkpath -f -m 0644 -o root:root "/var/log/${RC_SVCNAME}.log"
}

start() {
    while [ ! -S /var/run/docker.sock ]; do
        echo "Waiting for docker.sock to be available..."
        sleep 1.5
    done

    ebegin "Starting hawser"
    # We add $command_args_background here
    start-stop-daemon --start --background --make-pidfile \
        --pidfile "$pidfile" \
        $command_args_background \
        --exec "$command" -- $opts
    eend $?
}

stop() {
    ebegin "Stopping hawser"
    start-stop-daemon --stop --pidfile "$pidfile"
    eend $?
}

[riskwise99]

The log file says Error loading config: Docker socket not found at /var/run/docker.sock (sometimes more than once) even though the service file is depending on docker and is supposed to be starting after it.

Try an arbitrary delay, like 1 to 2 seconds, in a loop, once /var/run/docker.sock is met, off you go. That seems to be the key. No docker.sock, no hawser! LOL

But as you can see from my post, you're still not out of the woods and trees as hawser will crash and burn for no reason (for me at least!!)