Increasing jackd timeout seems to solve my issue.
Thanks Albert!
//Anders
2010/6/1 Anders Gunnarsson :
> Woups! Sorry for confusing things. I guess some messages in the
> conversation regarding this bug gave me the impression that it was
> Jack related. :-|
>
> It seems that bug is solved now. Yay!
>
> I'll try to increase the timeout to see if it helps me with my Jack
> issue. Still I think it's a bit odd that Mixxx is kicked when
> nozombies is used. I've never seen that with other clients, not even
> softsynths using most cpu, causing loads of xruns.
>
> //Anders
>
>
> 2010/6/1 Albert Santoni :
>> Anders: The bug the bounty is related to only involves problems when
>> you pick ALSA as the Sound API in Mixxx.
>>
>> If you're seeing problems with the JACK or PortAudio watchdog giving
>> Mixxx the boot, try running jackd with the --timeout 1 flag.
>>
>> (I usually compile PortAudio with --prefix=/usr, and it overwrites the
>> old version from the Ubuntu repo.)
>>
>> Thanks,
>> Albert
>>
>> On Mon, May 31, 2010 at 3:02 PM, Anders Gunnarsson wrote:
>>> I just want to check a few things to make sure I got everything right
>>> regarding this issue.
>>>
>>> If I build PortAudio from source. How do I know that this version is
>>> used before the installed binary? Right now I've used prefix /usr,
>>> which I guess would replace the binary install. If I use default
>>> /usr/local would that library override the binary insall in /usr?
>>>
>>> I've tried to build PortAudio r1411, but can't say I see any
>>> difference from any other version. It's kicked from jack with the
>>> error message below. I test on UbuntuStudio 10.04 64-bit with kernel
>>> 2.6.31-10-rt, and tried both Mixxx 1.7.1 from download page, and r2407
>>> from trunk.
>>>
>>> -jackd output
>>> Â subgraph starting at qjackctl timed out (subgraph_wait_fd=16, status
>>> = 0, state = Running, pollret = 0 revents = 0x0)
>>> alsa_pcm: xrun of at least 0.117 msecs
>>> bad status (1) for client PortAudio handling event (type = 5)
>>> -
>>>
>>> It seems PortAudio is kicked on the first xrun. Jackd is started with
>>> /usr/bin/jackd -R -Z -dalsa -dhw:1 -r44100 -p256 -n2. I guess -Z
>>> (nozombies) should make jack keep clients that are too slow, so this
>>> must be something else.
>>>
>>> //Anders
>>>
>>>
>>> 2010/5/29 Albert Santoni :
Thanks for the feedback guys! The bounty is up on Pledgie, if anyone
wants to add to the pot:
http://pledgie.com/campaigns/10964
Thanks again,
Albert
On Wed, May 26, 2010 at 9:12 PM, Albert Santoni wrote:
> Hi guys,
>
> It looks like a developer on the PortAudio mailing list has made a
> little bit of progress in troubleshooting our dreaded PortAudio xrun
> deadlock and thread termination problem:
> http://music.columbia.edu/pipermail/portaudio/2010-May/010234.html
>
> (For some info on this bug, see:
> https://bugs.launchpad.net/mixxx/+bug/383431 )
>
> As far as we know, this bug affects _every_ user on Ubuntu, and as a
> result, on our downloads page we recommend that Ubuntu users install
> an older version of PortAudio. Unfortunately, the users that get Mixxx
> from Synaptic or the Ubuntu Software Centre won't know that, and get
> Mixxx with a deadlock-prone version of PortAudio.
>
> None of us have ever dug into PortAudio's ALSA code seriously enough
> to figure out why the problem happens, probably because we've got
> enough bugs directly inside Mixxx to keep us busy. However, this bug
> is definitely causing problems for a huge number of people, and is
> probably severely affecting the growth of our userbase. We estimate
> roughly 50% of our users run Linux, and since most of those people are
> running Ubuntu, many people's first impressions of Mixxx are likely
> ruined by this bug. Because we get most of our developers from the
> Linux community, this bug needs to be fixed if we want to keep
> growing.
>
> None of the PortAudio maintainers have fixed this bug so far, and none
> of them are expressing any interest in doing it. Thus, we're faced
> with the following decision: Pull our own developer time away from
> Mixxx and try to convince one or more of us to fix the bug in
> PortAudio, or find a way of getting someone else to fix it. I work on
> Mixxx because it's super fun, but my enthusiasm for fixing PortAudio
> is very low, even though it's important to us. I think our entire team
> feels the same way, otherwise someone would have fixed the bug by now.
>
> However, I think this is a great opportunity to put our pool of funds
> that we've received from donations to good use. I'd like us to propose
> a $100 bounty on fixing this bug. It's not a huge amount of money, but
> I think it would provide a little bit of incentive for the guy on the
> PortAudio mailing list who's already started po
