On 6/22/19 12:16 PM, Adam Borowski wrote: > On Sat, Jun 22, 2019 at 11:26:40AM +0200, Tobias Hansen wrote: >> On 6/21/19 10:54 PM, Adam Borowski wrote: >>> (patch is >>> https://salsa.debian.org/pulseaudio-team/pulseaudio/merge_requests/5) >>> >>> On Fri, Jun 21, 2019 at 11:37:48AM +0200, Tobias Hansen wrote: >>>> I just updated by system from stretch to buster and after that there was >>>> no sound in GNOME because pulseaudio was not started. >>>> It can be easily worked around by setting "autospawn = yes" in >>>> ~/.config/pulse/client.conf but it's quite an annoying regression. >>>> >>>> Can this still be fixed for buster? Can we make it an RC bug? >> But I am using systemd and saw this regression anyway. Let me test that >> patch. > Then my patch can't possibly work for you, as it re-enables autospawn on > !systemd only. You may have ran into some other autospawn bug. > > In other words: > * on !systemd, PA never starts (which makes the package useless) > * on systemd, PA still fails to start in some cases > > I don't know enough about either PA or systemd to meaningfully help you > debug that case -- and my fix affects !systemd only.
Ok, my problem was that /run/user/1000 was mounted in the fstab of a schroot. That caused /run/user/1000 to be owned by root and then XDG_RUNTIME_DIR was not defined and systemd somehow didn't start pulseaudio. Sorry for the noise. Best, Tobias