[Asterisk-Users] * box hangs after a couple of days...

[Steven Critchfield]

On Tue, 2004-10-12 at 12:32 -0600, Michael Loftis wrote:
> 
> --On Tuesday, October 12, 2004 13:15 -0500 Steven Critchfield 
> <critch at basesys.com> wrote:
> 
> > You seem to have too much confidence in your raid setup. I'm wondering
> > if you are using a highpoint semi software raid since you have mentioned
> > an AMD machine and IDE drives. You probably are having IDE problems
> > arising from the number of interupts needed for zap hardware. Looks like
> > you need to go looking for some replacement hardware.
> 
> No I hate those highpoints.  Using Standard IDE, the Promise 'RAID' is 
> disabled.  We have a SCSI box here that's been purchased for the PBX 
> permanent home, but if this is how the Zaptel hardware makes a box behave, 
> interrupt loss causing severe disk corruption on a box that is sitting 
> around doing *NOTHING* then something is severely wrong either with the 
> hardware, or the drivers.  The box was entirely idle.  Only thing it's ever 
> done is sit, it's handled a total of 100 test calls as I've been setting it 
> up and screwed itself up pretty royally twice now.

Then you show how little you have learned about the hardware and the
system in general. ALL Zapata based hardware generates 1000 interupts a
second regardless of activity. It does so because a choice was made that
the complexity and cost involved in adding buffers and hardware to
service those buffers was too much. It was deemed that under most
situations, a PC is capable of handling that kind of load. 

So now we find out you are using the linux software raid to do IDE based
mirroring. Did you look into the preformance hit you take when mirroring
and worse when you eat the CPU with this performance hit so you can
write the data twice to the IDE bus. Expect this will be a problem if
the software raid will give up itself when zapata asks for an interupt.
If there is enough of these, and your system gets hung, your going to
have data loss to the journal and then to the rest of the drives. Worse
yet, if one drive had the updated info and it couldn't write to the
second drive, then you have trouble concerning keeping them both in
sync.  

<snipped>
I snipped this section just to keep from escalating the interaction to a
emotional level.
-- 
Steven Critchfield <critch at basesys.com>