RE: T2000 architecture reset on moderate-to-intense disk IO?

  • From: "Matthew Zito" <mzito@xxxxxxxxxxx>
  • To: <paulvallee@xxxxxxxxx>, <oracle-l@xxxxxxxxxxxxx>
  • Date: Fri, 30 Jun 2006 16:39:22 -0400

Odds are, this isn't a CRS-specific problem, but rather a problem with
the IO subsystem.  Solaris nodes dropping to an ok prompt can be caused
by an OS bug, but is more likely to be caused by some sort of odd
hardware-level event.  CRS, even on solaris, opts for rebooting a box
when it believes it needs to be fenced or dropped out of a cluster.
There's the Sun Verification and Testing Suite - VTS would be an
interesting experience to run for a few days.


From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Paul Vallee
Sent: Friday, June 30, 2006 4:31 PM
To: oracle-l@xxxxxxxxxxxxx
Subject: T2000 architecture reset on moderate-to-intense disk IO?

Hello all,
A new customer has an issue where their new T2000-based RAC cluster
running Oracle clusterware (CRS) has a node go to reset under moderate
to intense disk IO workload (10GR2, Solaris 10). The node that is being
connected to drops to the OK prompt and that's it. There are no logs or
messages in any logfile, CRS or otherwise. 
Does anyone have any similar experiences?

Paul Vallee
President, The Pythian Group, Inc.
I blog at 

Other related posts: