[pulseaudio-discuss] pulseaudio duplicate trace in /var/log/syslog

[Tanu Kaskinen]

On Sat, 2011-02-19 at 12:33 +0000, Jon Grant wrote:
> Hello
> 
> I have the following occurring every 5 seconds in syslog. I wanted to
> offer to submit a patch for review to remove some of the duplicate
> lines.

Lines 4 and 5 are not duplicates. If you look at the PIDs, there are two
pulseaudio processes. The daemon launching works so that the initial
process (29468) forks a new process (29470) that becomes the actual
daemon. The "failed to initialise daemon" message comes from the child
process, and it's useful information for debugging, since it indicates
that the execution reached that particular position in the code. The
"daemon startup failed" message comes from the parent process. It's
printed always if the parent process doesn't get a message from the
child process telling that startup was successful. It's not a good idea
to remove that message.

Line 3 looks quite redundant, though. But I checked the code, and the
place where it's printed is such that the message is printed whenever
loading the startup script fails for any reason, so just removing the
line probably isn't a good idea, otherwise other error situations may
not get logged adequately.

So, since there are no simple fixes, I don't think it makes sense to
spend effort to try to change the logging to be less repetitive. I can't
see this being a big problem. A bigger problem is that pulseaudio is
trying to start again and again, and always fails. The problem appears
to be that loading module-esound-protocol-unix fails, preventing the
whole daemon from starting. Apparently the esound socket is being used
by some other process. To find out which process, run

    netstat -l --unix -p | grep esd

> In addition, https://tango.0pointer.de does not have a signed SSL
> cert. Could it get a CAcert one? Or even a verisign SSL cert.
> Currently we get a firefox warning when connecting.

It's up to Lennart. When the web server was set up, I believe it was a
conscious decision to go with a self-signed cert, but I don't know if
the reasons for the decision are still valid today...

-- 
Tanu