giof71 / librespot-docker Goto Github PK

We are at a basic functionality for now.
It would be great to cover all the functionality of librespot through environment variables and volumes.

Avoid operations which require root if the container is run with a specified user other than root

As per the title

Hello,

I am trying to deploy this image in a Kubernetes cluster (K3s). The pod is running on a target node, which is a Raspberry Pi 3 with a HIFI-BERRY DAC+ plugged on it.

I have already tested several images but I always get the same error, it must be more related to the configuration than to the image itself.

I can connect via Spotify-connect, but after a few minutes the music stops and I get the following errors in the logs:

[2022-10-23T14:55:41Z ERROR librespot_playback::player] Unable to load audio item: MercuryError
[2022-10-23T14:55:41Z ERROR librespot_core::session] Connection reset by peer (os error 104)
[2022-10-23T14:55:41Z ERROR librespot_connect::spirc] subscription terminated
[2022-10-23T14:55:41Z WARN  librespot] Spirc shut down unexpectedly

And when I open a connection throught Spotify-connect, I very often get :

[2022-10-23T14:48:34Z WARN  librespot_playback::audio_backend::alsa] Error writing from AlsaSink buffer to PCM, trying to recover, ALSA function 'snd_pcm_writei' failed with error 'EPIPE: Broken pipe'
ALSA lib pcm.c:8545:(snd_pcm_recover) underrun occurred

I'm not sure if asound.conf is used in the image, but here's the configuration file I mount in the pod (/asound.conf):

pcm.!default {
  type hw card 0
}
ctl.!default {
  type hw card 0

And here are the environments variable provided to the container :

env:
  - name: DEVICE_NAME
    value: SPEAKER
  - name: DEVICE
    value: hw:0,0
  - name: DEVICE_TYPE
    value: speaker
  - name: FORMAT
    value: S24
  - name: BITRATE
    value: '320'
  - name: ENABLE_CACHE
    value: 'Y'
  - name: CACHE_SIZE_LIMIT
    value: 3G

I've already try to run without the cache. I suspect a PCM buffer problem but I would like your opinion ?

Thank you !

Update to latest github actions

I am wrongly using --normalisation-pre-gain, while the switch is --normalisation-pregain

We need to create docker images only when tags are pushed.
Sample tag names:

release/2022-01-01
devel/dev-snap-2022-01-01
feature/my-feature-snap-2022-01-01
main/main-snap-2022-01-01

Add quotes to command line at least to:

• ALSA_MIXER_DEVICE
• ALSA_MIXER_CONTROL

Then verify if other variables need to be quoted when transponed to the command line.

Is your feature request related to a problem? Please describe.
If we want an automatically starting container configured in PulseAudio mode on a desktop system, we probably cannot use a restart policy from docker, which seems to break PulseAudio (at least on my systems)

Describe the solution you'd like
As done with squeezelite-docker, a user level systemd service would solve the issue.

Describe alternatives you've considered
No alternatives that I know of, that do not require a manual interaction (like, e.g., starting an existing container from the command line or from Poirtainer)

Additional context
This applies to PulseAudio mode so basically to system with a desktop environment without auto-login. Automati login might probably mitigate the issue

The documentation does not explain how to use discovery mode (without credentials)

When using user mode, volumes should be chown-ed to the user we are impersonating.

• as per the title, support --device with values F64, F32, S32, S24, S24_3, S16. Defaults to S16.

Provide a few usage examples:

• docker run commands

• docker-compose.yaml files

• pulseaudio and alsa configurations

Avoid to rely on raspotify (a great project of course)
This currently does not work because cargo install seem to fail for armhf architecture on GitHub workflows.

• Parameters must be optional (when possible)
• Not specifying optional parameters should not break the container execution

• Documentation to be created

• allow selection of alsa device and other relevant parameters

On README.md in the Links section, librespot is mispelled as librepot

Enable reading credentials from a file.

Merge code for new release

Fix the workflow. A carriage return was removed

as per the title

Rebuild latest images after a long time.

Avoid leaving empty directories.
So we can delete $LIBRESPOT_PULSE_CONFIG_DIR/ if not empty (otherwise this can yield an error).

Perform the following changes:

• Remove apt proxy config
• Remove apt cache
• Create a second step with a scratch image, copy all data

Considering the fast release cycle of librespot, it would be nice if latest images are created on a daily basis