How to set ALSA default device to pulseaudio sound server on Docker?

Question

I want to play sound on Ubuntu 16.04 on Docker using PulseAudio and ALSA. However it outputs an error Device or resource busy. I think this is because the default device of ALSA is not set to the PulseAudio sound server (Note: it's set to the PulseAudio on the host Ubuntu). How can I change the default device to it? For example, can I do so by creating ~/.asoundrc or by modifying /etc/pulse/client.conf?

Steps to reproduce

$ # On the host Ubuntu...
$ aplay -L | head -n9
default
    Playback/recording through the PulseAudio sound server
null
    Discard all samples (playback) or generate zero samples (capture)
pulse
    PulseAudio Sound Server
sysdefault:CARD=PCH
    HDA Intel PCH, ALC295 Analog
    Default Audio Device
$ aplay /usr/share/sounds/alsa/Front_Center.wav 
Playing WAVE '/usr/share/sounds/alsa/Front_Center.wav' : Signed 16 bit Little Endian, Rate 48000 Hz, Mono
$ paplay /usr/share/sounds/alsa/Front_Center.wav  # Success

$ # On the Docker container...
$ sudo docker run -it --device /dev/snd ubuntu:16.04 /bin/bash
[email protected]:/# apt update
[email protected]:/# apt install alsa-base alsa-utils pulseaudio
[email protected]:/# aplay -L | head -n10
null
    Discard all samples (playback) or generate zero samples (capture)
pulse
    PulseAudio Sound Server
default:CARD=PCH
    HDA Intel PCH, ALC295 Analog
    Default Audio Device
sysdefault:CARD=PCH
    HDA Intel PCH, ALC295 Analog
    Default Audio Device
[email protected]:/# aplay /usr/share/sounds/alsa/Front_Center.wav 
ALSA lib pcm_dmix.c:1029:(snd_pcm_dmix_open) unable to open slave
aplay: main:722: audio open error: Device or resource busy
[email protected]:/# paplay /usr/share/sounds/alsa/Front_Center.wav 
Connection failure: Connection refused
pa_context_connect() failed: Connection refused

Environment

• Host: Ubuntu 17.04
• Docker: 17.09.0-ce

Additional notes

• On my environment, Jess's images don't work, too. Here is an example log.

  $ # When I play some sounds on host, `jess/spotify` fails to play music.
  $ sudo docker run -it \
      -v /tmp/.X11-unix:/tmp/.X11-unix \
      -e DISPLAY=unix$DISPLAY \
      --device /dev/snd \
      --name spotify \
      jess/spotify
  Gtk-Message: Failed to load module "canberra-gtk-module"
  libGL error: MESA-LOADER: failed to retrieve device information
  libGL error: Version 4 or later of flush extension not found
  libGL error: failed to load driver: i915
  libGL error: failed to open drm device: No such file or directory
  libGL error: failed to load driver: i965
  [1115/043835.051841:ERROR:sandbox_linux.cc(344)] InitializeSandbox() called with multiple threads in process gpu-process.
  [1115/043840.945653:ERROR:web_plugin_impl.cc(38)] Widevine registration is not supported after context initialization
  (spotify:1): GLib-GIO-CRITICAL **: g_dbus_connection_send_message: assertion 'G_IS_DBUS_CONNECTION (connection)' failed
  ALSA lib pcm_dmix.c:1052:(snd_pcm_dmix_open) unable to open slave
  

  The last line says that there is an error related to ALSA. (Note that jess/spotify can play music when there is no other software that plays sound. This may be because the sound device is not busy in such a case.)

• Although I self-answered this question, I still want more sophisticated solution or detailed explanation of the reason that some options are needed. Any comments would be appreciated. Thank you.

Solution

Thanks to fsmunoz's this comment, I can play sounds on the container using pulseaudio, without modifying any config files. The points are the following:

• Set PULSE_SERVER in order to let container's pulseaudio know the server address.
• Add the container into the audio group.
• Share pulseaudio's cookie.

And here is a working example:

$ sudo docker run -it \
    --device /dev/snd \
    -e PULSE_SERVER=unix:${XDG_RUNTIME_DIR}/pulse/native \
    -v ${XDG_RUNTIME_DIR}/pulse/native:${XDG_RUNTIME_DIR}/pulse/native \
    -v ~/.config/pulse/cookie:/root/.config/pulse/cookie \
    --group-add $(getent group audio | cut -d: -f3) \
    ubuntu:16.04 /bin/bash
[email protected]:/# apt update
[email protected]:/# apt install alsa-base alsa-utils pulseaudio
[email protected]:/# aplay -L | head -n9
default
    Playback/recording through the PulseAudio sound server
null
    Discard all samples (playback) or generate zero samples (capture)
pulse
    PulseAudio Sound Server
sysdefault:CARD=PCH
    HDA Intel PCH, ALC295 Analog
    Default Audio Device
[email protected]:/# aplay /usr/share/sounds/alsa/Front_Center.wav 
Playing WAVE '/usr/share/sounds/alsa/Front_Center.wav' : Signed 16 bit Little Endian, Rate 48000 Hz, Mono
[email protected]:/# paplay /usr/share/sounds/alsa/Front_Center.wav  # Success!

Note: Still, I don't know why the cookie is needed. Without the cookie, an error ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect: Access denied occurred.

Comments

[Socrates]

Try this: stackoverflow.com/questions/41083436/…

[nekketsuuu]

@Socrates Thanks for your suggestion, but I already tried it. The key of that answer is to use --device /dev/snd.

[Web User]

Can I achieve a similar goal with a Ubuntu Docker container running on a Windows 10 host?

[nekketsuuu]

@WebUser No. I don't know the details but some configuration, such as file paths, must be different. In addition, you should install PulseAudio to Windows if you want to follow this answer. I recommend that you ask a new question about the configuration.

[LIH]

In Ubuntu server 20.04 I get this error: WARNING: The XDG_RUNTIME_DIR variable is not set. Defaulting to a blank string.