PERC3/Di failure workaround hypothesis

Sean Bruno - TELECOM sean.bruno at metro1.com
Sat May 29 16:19:00 CDT 2004


Let me clarify a bit.

The 80+ machines I have that ARE NOT experiencing any of these problems
are running RH 7.2 and 7.3.  They are all utilizing the 3Di in a RAID1
configuration.  They all have two 32GB drives.  Most, if not all, are
running BIOS version A10.
The 3Di is running version the 2.7 version on all of my machines.
These machines have NEVER exhibited the failure that is described in
this thread.

The machine that I am testing on, is running AS3.0 with BIOS version A17
and the 2.7 version of the 3Di "firmware."  It is surviving about 3 to 4
days in between implosions.  The drives are in a RAID 1 configuration
and are 32GB in size each.
I have completely re-installed the system, rebuilt the container from
the 3di bios menu, and disable all caching.  It blew up yesterday around
noon PST.

What conclusions can you draw from this information?  What would you
like me to do next?

On Sat, 2004-05-29 at 12:59, Matt Domsch wrote:
> On Sat, May 29, 2004 at 09:54:55AM -0700, Sean Bruno - TELECOM wrote:
> > Test results are in.
> > 
> > My test machine with the read/write cache disabled "detonated" last
> > night after about one day of up-time.
> > 
> > Looking at the console, I see a smattering of errors writing to the
> > disks.  
> > 
> > What seems very strange to me is the fact that I have close to 80 2650's
> > running just fine with RAID1.  These machines are a bit out of date as
> > far as firmware is concerned.
> > 
> > What do you want to try next?
> 
> OK, first we need to know what those messages are.  Serial consoles
> are your friend here, if no other way.
> 
> Then we need to know what firmware you're running.  If as you say,
> it's out of date, if it's less than 2.7, please upgrade to 2.8 and try
> again.  Other bugs have been fixed in the firmwares which could
> account for this too, where the 'disable cache' workaround wouldn't
> fix.
> 
> If that still doesn't resolve it, report back here for specific
> instructions for calling Dell Tech Support.
> 
> Thanks,
> Matt
-- 
Sean Bruno - TELECOM <sean.bruno at metro1.com>




More information about the Linux-PowerEdge mailing list