Help - Oracle 9i RAC on HP ServiceGuard unstable

[Stephen Andreassend]

Hi,

We are trying out the HP cluster platform for Oracle 9i RAC, and our HP consultants are struggling to get it configured properly.

When testing to see what happens when 1 of the 2 redundant HyperFabric/2 links goes down, the entire cluster stops.

When disk connectivity is lost on one node, all nodes in the cluster hang and all Oracle sessions freeze.

We do not have any HP consultants in this country that know how to configure Oracle/ServiceGuard with EMS to handle hardware failures.

If you run OPS or RAC on HP Service Guard, can you please send me your MC/SG+packages+Oracle+EMS config files (and any other tips) so I can see what you do to ensure the cluster remains available during a hardware failure. At this stage our HP cluster does not deliver any of the promised High Availability.

Thx
Steve

[Charles McCary]

Stephen,

when you say that the cluster stops, can you be more specific? Do the packages halt? Do the machines panic?

tx,
c

[Stephen Andreassend]

Another example, if 1 RAC node is rebooted, this is what happens on the surviving node.

$ sqlplus system/manager@RAC

SQL*Plus: Release 9.0.1.3.0 - Production on Thu May 9 00:45:53 2002

(c) Copyright 2001 Oracle Corporation. All rights reserved.

ERROR:
ORA-27500: inter-instance IPC error
ORA-27300: OS system dependent operation:HMP is not UP failed with status: 4
ORA-27301: OS failure message: Interrupted system call
ORA-27302: failure occurred at: skcsicini

[Stephen Andreassend]

The above error goes away when the rebooted node rejoins the cluster.

[Stephen Andreassend]

Is it possible to tune/configure the HMP protocol just like TCP?

Where? How? Where are the docs on this?