Jackd Problems -- alsa_pcm: xrun of at least 1227061150613.504 msecs

[Jonathan Ryshpan]

On Thu, 2008-11-20 at 11:16 +0100, Jeremy wrote: 
> Hello Jonathan,
> 
> To get jack running properly you need:
> - Planet CCRMA packages 
> (http://ccrma.stanford.edu/planetccrma/software/) which are not 
> available for F9
> - low latency kernel, Planet CCRMA also provides these, unfortunately 
> not for F9
> - zero latency soundcard, an onboard card probably never works without 
> an xrun every now and then
> 
> An xrun is indeed a buffer under/overrun and could be caused by numerous 
> things, always hard to put a finger on it. Could be software, hardware. 
> Best is to run jack with a light weight WM like fluxbox or IceWM and to 
> use as little other programs as possible. And to switch to another 
> version of Fedora (like 7 or 8) for which there are Planet CCRMA 
> packages available, otherwise it's no use I think.

Thanks for the info.  The fancy kernels, etc. are probably not necessary
to avoid xruns in a properly configured Fedora system with a strong
enough CPU, etc.  My own CPU is AMD x86_64 running at about 1 GHz.  In
the past I have had typically 1 xrun per hour, which hasn't affected the
sound quality of recordings perceptibly.

However your message doesn't answer the question: what does the time
interval associated with an xrun mean?  Do you (or anyone else) know?  A
reply to this question posted on the Jack-devel list indicates that the
very long time interval here is the result of a bug in this version of
jackd:
        MarcO'Chapeau <marco at marcochapeau.org> writes: 
        That's a bug fixed after 0.109.2 I'd say (is that still only in SVN ?). It
        was due to a comparison of 2 timestamps not coming from the same clock
        source afair
I have only one clock afaik on my system but who knows...

An attempt to build jackd on my system has failed due to problems in
portaudio as released in the F9 distro.  More on this in my next message
to the list.

Thanks - jon