[Linux-cluster] DRAC5 fencing problem with Redhat 5.3 cluster

[Marek 'marx' Grác]

Dan Hayes wrote:
> I have 8 Dell servers all loaded with Redhat 5.3.  The machines have 
> an external and internal IP address.  The cluster is configured on the 
> internal ip address.
>
> The cluster appears to be working fine except for fencing.  I try to 
> manually fence a node, but get a connection error
> [root at w002 ~]# fence_node lb001.domain.com <http://lb001.domain.com>
> agent "fence_drac5" reports: Unable to connect/login to fencing device
>
> I was able to get the fence_drac5 command to work directly by adding 
> the "-x" option for using SSH.  It says the connection times out, but 
> it does shut down the other machine
> [root at w002 ~]# fence_drac5 -a 165.289.178.221 -l root -p pass -x 
> lb001.domain.com <http://lb001.domain.com>
> Connection timed out
>
> Why does the connection time out?  And how do I add the "-x" option to 
> the cluster.conf file so that the fencing agent can connect?
Can you send me verbose output (-v) ? It is possible that the machine 
needs more time than POWER_TIMEOUT (20 seconds) or output of status is 
not parsed correctly ('Starting machine' instead of ON/OFF).

-x is equal to option secure="1" (we don't care about inserted value)

m,