One minute after one machine in a cluster panics where CPS fencing is used, the 2nd node panics on Infoscale 8.0.2.2700/RedHat 9.4

book

Article ID: 1000766271

calendar_today

Updated On:

Description

The first panic happened as a result of a network partition whereby a vxfen-induced panic took place. However, a minute later the 2nd node in the cluster panicked.

Cause

Issue related to GAB handling the departure of the 1st node when it panics. It reboots faster than anticipated by the surviving node and in its handling/clean-up procedure ends up firing a GAB iofence and panics the surviving node.

Resolution

A hotfix (GAB 8.0.2.2801) was provided to remedy the GAB defect and prevent the 2nd panic in this scenario. Please refer to Technical Support if requiring the hotfix.

Issue/Introduction

One minute after one machine in a cluster panics where CPS fencing is used, the 2nd node panics on Infoscale 8.0.2.2700/RedHat 9.4