Proper way for a server to spawn Pulseaudio?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, 2016-08-31 at 23:03 +0200, Espen Jürgensen wrote:
> Dear all, first of all thanks for your work on the Pulseaudio project!
> 
> I maintain a Linux music server, and I am working on adding a Pulseaudio
> output for it. This is for local output. So far it has been simple and easy
> working with Pulseaudio, but going ahead I am unsure about the best way of
> spawning the Pulseaudio daemon. I hope someone here can give advice.
> 
> Most users of the music server will have it running as either root or a
> dedicated user, and will run it on a headless server without any desktop
> environment. Could be an RPi, for instance. So pretty much like mpd.
> 
> In such a setup the music server currently gets "Connection refused",
> unless Pulseaudio has manually been started in system mode by the user (I
> think...?). So, I have googled what to do in that situation, and that has
> left me a little a confused. For instance, some of the mpd guides say that
> Pulseaudio should autospawn - but that doesn't seem to happen for me, so
> perhaps there is some condition that needs to be met? I have, of course,
> set autospawn = yes (was already the default).

Autospawning works only in the per-user mode. In the system mode the
expectation is to have an init script or a systemd service that starts
pulseaudio at boot (with systemd, socket activation is possible too).

We don't provide init scripts or systemd service file for the system
mode. Init scripts can't be provided, because they are not portable
across distributions. Shipping a systemd service file would probably
make sense, though.

> Other guides tell you to use TCP and add "auth-ip-acl=127.0.0.1" in
> default.pa - but that seems a bit clumsy? Then there is also
> recommendations on adding the server user to the Pulseaudio groups (pulse,
> pulse-access), which I have tried, but which didn't seem to change much.

With the default settings, all users that want to use pulseaudio when
it's running in the system mode have to be in the pulse-access group.
The pulse group is only meant for the pulseaudio daemon - don't add
users to that group.

> Of course, I would like to inconvenience the user as little as possible
> with configuration activities, so with that in mind my question is: What is
> the proper way to make a server spawn Pulseaudio, given that it is running
> as root or dedicated user and that there is no desktop?

I think starting pulseaudio at boot in the system mode makes most
sense for headless systems. I don't think another server-type
application should spawn it.

When your server runs under a dedicated user, autospawning should work,
but then you'll probably run into problems with not having access to
the audio hardware, because typically the device permissions are
handled dynamically so that only the currently-active login session has
access. Trying to "fix" this by making the device permissions static
will cause breakage, because your music server will then hog the
hardware so that regular users can't access it anymore. It's probably
best if you use the PA_CONTEXT_NOAUTOSPAWN flag in your application
when connecting to pulseaudio, unless your application runs inside a
regular user login session.

-- 
Tanu


[Index of Archives]     [Linux Audio Users]     [AMD Graphics]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux