[LAU] Jack conking out, How to debug?

[Paul Davis]

On Thu, Feb 17, 2011 at 1:13 PM, Joel Roth <joelz at pobox.com> wrote:
> On Thu, Feb 17, 2011 at 01:09:10PM -0500, Paul Davis wrote:
>> On Thu, Feb 17, 2011 at 12:50 PM, S. Massy <lists at smassy.andropov.org> wrote:
>> > On Wed, Feb 16, 2011 at 02:35:43PM -0500, Paul Davis wrote:
>>
>> >> why are you using -t100 ?
>> > A leftover from my non-RT kernel, but I gigured it wasn't bad to give
>> > processes an extra grace period. Can this have negative side-effects in
>> > any way?
>>
>> -t 100 isn't "extra grace period". it overrides however long JACK
>> would have waited (typically fractionally less than one period).
>
> According to the man page for jackd 1.9.6, the default value
> for -t is 500ms.

I didn't explain that very well. JACK will wait for "the client
timeout". By default its 500msec. But specifying -t 100 doesn't add
100msecs to that, it means that JACK will wait only 1/5 as long for
clients before concluding that they are not coming back.