[Asterisk-Users] Interrupt latency problems

[Rich Adamson]

> > > I'm debugging a TxFax problem whereby the fax transmission fails.  I 
> > > suspect interrupt latency--some interrupt routine is holding its 
> > > interrupt too long.  I have all unnecessary services switched off and X 
> > > is not running when I perform these tests.  Some transmission are 
> > > successful while others fail at random points.
> > > 
> > > I've noticed that after I boot Linux, load zaptel, wcfxo, and wcfxs, and 
> > > then run System Monitor, the CPU utilization spikes to 60% every 10 
> > > seconds.  After I remove wcfxs the spikes stop, and system utilization 
> > > stays at 0%.
> > > 
> > > I have two TDM22B cards and get these spikes with either card in the system.
> > > 
> > > This is an Athlon-64 3000+ CPU and Fedora Core 2.  zaptel-1.0.2.
> > > 
> > > Is there anyone who has a TDM card on a "quiet" system who can run 
> > > System Monitor and observe CPU utilization?
> > > 
> > > Thanks for your help.
> > > 
> > > P.S.  I do 'hdparm -u1 /dev/hdx' on the disks.
> > > 
> > > 
> > 
> > I have a athlon xp 1600 with a gig of memory a tdm40 with 2 fxs ports and a 
> > t100p t1 hooked to a audiocoded tp260 and see minimal cpu utilization when idle.
> > 
> > run top and see what is using the cpu.
> 
> top will only report userspace problems, and to top it off, top only
> reports on snapshots of the system on as low as 1 second. With Zap
> hardware hitting the system 1000 times a second for service, you might
> happen to get an occasional hit time here top and the hardware hit
> pretty close to show extra load. top also has the problem of effecting
> the system it is watching. It is a lot like those pesky physics problems
> where what you use to measure changes the object your measuring.
> 
> Basically all that is to say that top probably won't tell you what you
> want to know.About the only thing that would be of interest is if the
> percentage is viewed in the system or userspace portions. If in system,
> you will have to go debugging the kernel. 
> 
> Of course, it seems this is mostly being reported against RH and FC.
> I'll take a quick guess that it isn't bad users so it would leave you
> with bad kernels. My personal opinion is to not trust what the distros
> do to the kernels. Even in my beloved Debian I don't trust the default
> kernel. I suggest you download a stock vanila kernel from kernel.org and
> config it as minimally as possible for your hardware and try and see if
> it reproduces the problems you are seeing currently. 

Critch,

Help me understand the above. When running top without any parameters, I
see the 30% cpu jump. Killing asterisk, I see the same. Killing zaptel,
the 30% goes away. Restarting zaptel, the 30% comes back.

That would suggest that zaptel (and wctdm in my case) are impacting the
30% cpu consumption (and the fax stuff for the original poster). The 1,000
interrupts are going on constantly, so should not have any impact on top's
longer term sampling. If I change top to 1 second samples, I then see
100% utilization every ten seconds. That only implies that top is averaging
the numbers over the sampling interval (for cpu utilization).

So, isn't the issue he/I are chasing after essentially 'why is cpu consumption
jumping 30% (or 100%) every ten seconds when zaptel is running with
no calls present?

Rich