[Linux-cluster] fence_apc_snmp woes

Jeremy Carroll Jeremyc at tasconline.com
Fri Aug 24 13:35:55 UTC 2007

Try to run the fence agent manually. See if it works. Most likely the
problem is fenced cannot successfully run the program to power down the
switch. Post your results.

fence_apc_snmp -v -a oc-index3 -c <passwd> -n 18 -o off 

-----Original Message-----
From: linux-cluster-bounces at redhat.com
[mailto:linux-cluster-bounces at redhat.com] On Behalf Of Brian Sheets
Sent: Friday, August 24, 2007 6:06 AM
To: linux clustering
Subject: [Linux-cluster] fence_apc_snmp woes

I have a 2 node cluster on debian. Below is my cluster.conf. If I down
node1's nic 
node2 sees and tries to fence

Aug 24 10:57:38 oc-index4 fenced[7599]: oc-index3 not a cluster member
after 0 sec post_fail_delay
Aug 24 10:57:38 oc-index4 fenced[7599]: fencing node "oc-index3"
Aug 24 10:57:38 oc-index4 fence_manual: Node needs to be
reset before recovery can procede.  Waiting for to rejoin
the cluster or for manual acknowledgement that it has been reset (i.e.
fence_ack_manual -n
Aug 24 10:59:34 oc-index4 fenced[7599]: fence "oc-index3" success

It states that it's fencing, but never does, and if I do a
fence_ack_manual, then fence_apc_snmp gets run and the node1 gets
powered down.

what am I missing?

<?xml version="1.0"?>
<cluster name="index" config_version="2">
<cman two_node="1" expected_votes="1">
<clusternode name="oc-index3" votes="1">
                <method name="single">
                       <device name="oc-cab1-pdu2" port="18"

<clusternode name="oc-index4" votes="1">
                <method name="single">
                  <device name="oc-cab1-pdu1" port="16" option="off"/>

        <fencedevice name="oc-cab1-pdu2" agent="fence_apc_snmp"
ipaddr="" login="apc" passwd="xxxx"/>
        <fencedevice name="oc-cab1-pdu1" agent="fence_apc_snmp"
ipaddr="" login="apc" passwd="xxxx"/>

Linux-cluster mailing list
Linux-cluster at redhat.com

More information about the Linux-cluster mailing list