Since I had to have a working patch by today I was changing the problematic patch into one big one containing all the other patches that were to be started at startup of the first patch. This makes it kind of inflexible for now but: The "watchdog signalling pd..." is gone.
I'm still not sure if the errors in the patch were caused by loading problems with watchdog caused but since I've changed it any strange behaviour has disappeared.
Generally, as soon as I start loading several (large) patches at the same time in "-nogui" mode this happens. No problem with the graphics running.
I think I'll simply avoid loading more the one patch at the time just to ensure things are loading correctly.
Since I'm not a programmer and really good with all that source code and compiling stuff I'll keep my hands of that to avoid even more errors than I had before.
If anybody wants to investigate more it would be great but I'm kind of in a time pressure and have to get the patch I'm working on running as fast and safe as possible.
Ingo
Sorry, I just tried and killing the watchdog kills pd too, oops.
Only way I can see to do it is comment out the line in s_watchdog.c:
kill(getppid(), SIGHUP);and recompile.
cheers Miller
On Tue, Nov 17, 2009 at 07:50:31PM +0100, Ingo Scherzinger wrote:
Freezing the machine shouldn't be a problem. It's only running one pd
patch
which is being automatically started right after the system (Ubuntu
Stutio
8.04) has booted up. The patch won't be edited after that. I'll have to see how I can kill watchdog before it does anything. Is
there a
startup flag I could use to start pd without watchdog or what would be
the
best way to start pd without it?
Ingo
Hmmm -- it shouldn't have any effect, but I can't swear i doesn't. To find out, I think you can just kill the watchdog process and see if the flakiness goes away. Only downside to that is you no longer have protection against Pd freezing your machine :)
Miller
On Tue, Nov 17, 2009 at 05:54:11PM +0100, Ingo Scherzinger wrote:
I am getting pd-watchdog signals during loading of a pd patch on a
"-
nogui"
system.
I read that watchdog is trying to stop the pd thread to prevent
lockup
when
it can't get any pings to respond in time. Now the question is:
Is it possible that there are portions of the pd patch not loading
correctly when this happens? BTW it happens with the "-rt" flag on
and
off.
And the patch is very large.
I just noticed the reason this might be happening is that I am
starting
some
other patches (samplervoices) during the initial loading of the
patch
which
need quite some time to load. But the question remains: can there be
loading
errors because of that? And I do have some strange things happening
with
this patch lately!
Ingo
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management ->