Novell Cluster Services can not be unloaded when using Bonding

  • 7001097
  • 05-Aug-2008
  • 30-Apr-2012

Environment

Novell Open Enterprise Server 2 (OES 2)
Novell Open Enterprise Server (Linux based)

Situation

Environment:
- SUSE Linux Enterprise 10 Support Pack 1 + Patches
- Open Enterprise Server 2
- Novell Cluster Services

Problem:
- When shutting down the *last* node in a OES Linux Cluster environment, that last node hangs in the process.
- When using bonding, Novell Cluster Services can not be unloaded. It happens when you try to use rcnovell-ncs stop, rcnovell-ncs restart
or in a crash/failover scenario.

Symptoms:
- modprobe command goes to 100% cpu load when it can not unload gipc.
- the "modprobe -r" command in the "uldncs" script never completes and gets stuck deleting the first kernel module (cma)
- if this node is the only cluster node, or the only remaining alive cluster node.


Resolution

Replace /lib/modules/kernel_version_used/updates/ncs/gipc.ko module with a newer version and restart  your cluster.

This fix is already available at Bug#411156 and it should be available in the update channel soon.