[Asterisk-Users] RAID affecting X100P performance...

[Mike Benoit]

On Thu, 2004-07-22 at 12:56 +1200, wrote:
> > I would be interested in seeing if other people can reproduce low
> > zttest accuracy rates with their mainboards. zttest is in the zaptel/
> > directory, and you can run it while Asterisk happily chugs along
> > handling calls.
> > 
> > What I usually do is run zttest in one window, then in another window
> > run "updatedb" or "find / > /dev/null" or "hdparm -t /dev/hda" or all
> > of them at the same time. The harder and longer you work the drives
> > the better. If the problem exists on your system, you'll see the
> > zttest accuracy drop below 99.98% at some points.
> > 
> Okay, I have 1 X100P and 4FXS modules (TDM400) and running zttool 
> under normal load gets about 99.98.
> 
> I have 1 hard disk, no raid.
> 
> If however I run updatedb on another console I get drops down to 
> 33.508301, but still spends most of the time between 99 and 100.
> 
> I'll have to remember not to run updatedb during business hours 
> (actually it's run from cron in the middle of the night anyway).
> 
> What other task would an asterisk system do that would use so much 
> disk though?  Maybe recording every conversation on a fully loaded 
> quad T1 box...
> 

It's not just that, I've found I'll get "cutting out" even when the disk
isn't really being used. The test seems to show that your mainboard
can't handle the load. I'm still not 100% sure every mainboard doesn't
do the same thing, but I couldn't re-create it on a brand new Celeron
2.4ghz box. The Celeron was the only system I had that I couldn't get
the accuracy to drop below 99.98%, a older P3-800, and a older Celeron
533 all had major issues.

Have you had any complaints of "cutting out"? It can be _very_ minor,
and usually not noticeable during normal conversations. In the minor
cases it may sound like just parts of words "go missing" but the person
can still make out the word in the end. Also I have a hunch it _may_ be
contributing to DTMF tones not being recognized in rare cases. I've had
the odd complaint that people have to sometimes press buttons twice
before my menu will "clue in".

Try being on a call that traverses the X100P card, or the TDM400 for
that matter I'm interested if it has the same issue, and run the same
test. See if you notice any "cutting out". 

-- 
Mike Benoit <ipso at snappymail.ca>