[Pkg-tigervnc-devel] Bug#1013979: tigervnc-standalone-server: tigervncserver listening on UNIX sockets doesn't start session and dies

[Christoph Anton Mitterer]

Package: tigervnc-standalone-server
Version: 1.11.0+dfsg-2
Severity: normal


Hi.

I've had reported this originally upstream at:
https://github.com/TigerVNC/tigervnc/issues/1485
but didn't notice that `tigervncserver` was a Debian addition.

So I'm more or less copy&pasting the information from there...


When running tigervncserver with listening on UNIX sockets
and while it simply waits for the first client to connect
(i.e. no attempt to connect is made yet), it dies after
-ClientWaitTimeMillis ms.

The reasons seems to be, that as soon as I add -rfbunixpath
no session is started, thus no xterm whatsoever is invoked either.

Even when I manually specify a session command like in:
  tigervncserver -fg -localhost=yes -nolisten=tcp -rfbunixpath=/home/calestyo/remsock :5 -- /usr/bin/xterm
or
  tigervncserver -fg -localhost=yes -nolisten=tcp -rfbunixpath=/home/calestyo/remsock :5 -- /etc/X11/Xtigervnc-session
nothing happens (i.e. xterm nor the session get started).


Steps to reproduce the behavior:
1. tigervncserver -fg -localhost yes -nolisten tcp -rfbunixpath /tmp/someSocket
2. wait 30 s (default of -ClientWaitTimeMillis is 30000 ms)
3. it dies


When I manually start e.g. DISPLAY=:5 xman then it seems to work.
Interestingly, when subsequentially killing xman, the vnc server
stays alive,... even after the 30s, which AFAIU the description
of -ClientWaitTimeMillis shouldn't be the case either.

But it's still a bit strange that it uses -ClientWaitTimeMillis at
all, because that shouldn't have to do with it?
Maybe just a coincidence or maybe tigervncserver misues that setting?


Any ideas?
Thanks,
Chris.