Add Dockerfile

[jtabet]

This pull requests adds a way to easily integrate gmrender-resurrect in a docker environment. The "FRIENDLY_NAME" environment variable must be provided, and additional volume mounts for pulseaudio.

Here is what's needed to use pulseaudio in docker-compose:

    volumes:
      - /etc/localtime:/etc/localtime:ro
      - $HOME/.config/pulse/cookie:/root/.config/pulse/cookie
      - ${XDG_RUNTIME_DIR}/pulse/native:${XDG_RUNTIME_DIR}/pulse/native
    environment:
      - PULSE_SERVER=unix:${XDG_RUNTIME_DIR}/pulse/native
    devices:
      - "/dev/snd:/dev/snd"
    group_add:
      - audio

[mill1000]

Cool. Just a few thoughts from me.

• What about using Alpine as the base image? It's generally smaller than Debian images.
• FRIENDLY_NAME shouldn't be required. I would either update the command to conditionally use the variable, or have a default value.
• UUID should also be a runtime option to enable the same image to be deployed multiple times.
• An environment variable should be added to allow specifying arbitrary command line arguments. (e.g. output options, MIME filters)
• Environment variables are not declared with an ENV instruction.
• What about ALSA instead of PulseAudio? Even better if its a build option
• In my experience it's better to use a valid init for the entrypoint and run the service as a command. I like to use tini.

Here's an untested Dockerfile that I whipped up. It builds both backends but I haven't confirmed if I can render anything.

ARG AUDIO_BACKEND="alsa"

# Build image
FROM alpine:latest AS build
ARG UUID
RUN apk add --update build-base autoconf automake libtool pkgconfig gstreamer-dev libupnp-dev uuidgen
WORKDIR /opt/gmrender-resurrect
COPY . .
RUN ./autogen.sh && ./configure CPPFLAGS="-DGMRENDER_UUID='\"${UUID:-`uuidgen`}\"'"
RUN make && make install DESTDIR=/gmrender-install

# ALSA image
FROM alpine:latest AS alpine-alsa
RUN apk add --update alsa-lib alsa-utils

# PulsaAudio image
FROM alpine:latest AS alpine-pulse
RUN apk add --update pulseaudio

# Run image
FROM alpine-${AUDIO_BACKEND}
COPY --from=build /gmrender-install /
RUN apk add --update tini libupnp gstreamer gstreamer-tools gst-plugins-base gst-plugins-good gst-plugins-bad gst-plugins-ugly
ENV FRIENDLY_NAME=
ENV UUID=
ENV OPTIONS=
EXPOSE 49494
ENTRYPOINT ["/sbin/tini", "--"]
CMD ["/bin/sh", "-c", "/usr/local/bin/gmediarender --logfile=stdout ${FRIENDLY_NAME:+-f \"$FRIENDLY_NAME\"} ${UUID:+--uuid \"$UUID\"} $OPTIONS"]

It defaults to an ALSA backend but PulseAudio can be build with:

docker build -t gmrender_pulse:latest . --build-arg AUDIO_BACKEND=pulse

Here's a comparison of image size

gmrender_debian         latest    7dc5e33f3a94   19 seconds ago      791MB
gmrender-alsa           latest    27c29f38b254   3 minutes ago       160MB
gmrender-pulse          latest    2d4c4b6b8525   4 minutes ago       161MB

[jtabet]

I confirm it builds and runs with your changes. However, I cannot provide in depth tests, since I stopped using gmrender-resurrect because I can't use UPnP in my project.

[mill1000]

I've confirmed the ALSA backend works on a Raspberry Pi 3 with the following Docker Compose.

services:
  gmrender:
    image: gmrender:latest
    container_name: gmrender-resurrect
    restart: unless-stopped
    network_mode: host
    environment:
      OPTIONS: "--mime-filter audio"
    devices:
      - "/dev/snd:/dev/snd"