Cman configures cluster communication on wrong interface on Red Hat Enterprise Linux 6
Issue
-
We got cluster failure in our RHCS yesterday, it seems to be fence race ( or loop ). Our configuration is:
- bond2 : Service network
- bond1 : Heartbeat Network
- eth2 : Dedicated Network
-
We unplugged cables of eth2 on both cluster nodes. After unplugged cables of eth2 on both, system had rebooted repeatedly by fence loop ( race ) between two nodes.
- eth2 is not network related to cluster, why fence loop ( race ) occured?
Environment
- Red Hat Enterprise Linux Server 6 (with the High Availability or Resilient Storage Add Ons)
- Red Hat High Availability cluster with 2 or more nodes
-
Clusternode name
entries in/etc/cluster/cluster.conf
do not matchuname -n
output:- Clusternode name from cluster.conf:
$ grep "clusternode name" /etc/cluster/cluster.conf <clusternode name="node1_hb" nodeid="1"> <clusternode name="node2_hb" nodeid="2">
uname -n output
:
$ uname -n node1
- There is no
NODENAME
override configured in/etc/sysconfig/cluster
or/etc/sysconfig/cman
.
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.