[winswitch] pulse audio processes on server remain after xpra stop
antoine at nagafix.co.uk
Fri Jun 12 21:22:44 BST 2015
On 12/06/15 22:03, Jiang Qian wrote:
> That's fine. I'll just kill it manually by kill -9. There's no need to
> fix it then.
0.14.26 and 0.15.1 will include the fix for pulseaudio under Ubuntu, I
am in the process of testing the packages before the release, so you can
find them here already:
(though if we can fix your chrome DPI issue quickly, I'll delay the
release to include this fix)
> Thanks for looking into this. My troubles right now is more with the
> latest update with chrome. See ticket #877.
You mean #887:
> I wonder if anyone else has the issue, or is this another ubuntu problem.
Chrome is generally problematic (client side decorations in general),
though we are working on improving that.
> On Fri, Jun 12, 2015 at 9:24 AM, Antoine Martin <antoine at nagafix.co.uk> wrote:
>> On 11/06/15 21:50, Jiang Qian wrote:
>>> Hi everyone,
>>> I'm puzzled whether this behavior is intended or not, so I ask this
>>> list rather than file a bug report. When I quite xpra instance on the
>>> server by doing first
>>> xpra list
>>> to find out which display it's running on (e.g. :100)
>>> and then do
>>> xpra stop :100
>>> the xpra instance stops successfully, as xpra list shows no running instance.
>>> But if I do this on the server
>>> ps|grep pulse
>>> and I found instances of pulse audio stopped by xpra still running.
>> That would be an Ubuntu bug, I have created a ticket for it:
>>> When I start a new instance of xpra server, a new instance of pulse
>>> audio start. If I kill the old pulse audio, the sound on the new xpra
>>> instance is not affected.
>> Just be aware that you can only have a single pulseaudio instance per
>> user unless you start messing with its environment.
>> That's a pulseaudio "feature".
>>> This happens both on 0.14 branch and 0.15 branch. I'm running both
>>> server and client on ubuntu 14.04
>> I've just tried this with a brand new user and the pulseaudio process
>> did get killed when the server was stopped. But that was on Fedora...
>> In your server log, you should see messages like these right at the end:
>> Shutting down in response to request
>> stopping pulseaudio with pid 19682
>> xpra is terminating.
>> removing socket /home/xpratest/.xpra/myhostname-20
>> killing xvfb with pid 19665
>> You can also see details on the processes started by xpra with:
>> xpra info | grep "^child"
>> And then I tried it on Ubuntu Trusty and found that it restarts a brand
>> new process after we kill the pulseaudio server.
>> So, it's a "feature" that sending a SIGINT does not stop it properly, sigh.
>>> shifter-users mailing list
>>> shifter-users at lists.devloop.org.uk
>> shifter-users mailing list
>> shifter-users at lists.devloop.org.uk
More information about the shifter-users