Red Hat Bugzilla – Bug 164627
fence_ipmilan does not have a default operation; should be "reboot".
Last modified: 2009-04-16 16:17:55 EDT
Description of problem:
Created attachment 117273 [details]
Created attachment 117274 [details]
Fixes problem in both cases (stdin + cmdline)
*** Bug 164997 has been marked as a duplicate of this bug. ***
Note that this bug means that the ipmilan method is unusable in production, as
fence_node doesn't work on those devices.
Note that the error message indicates that you need to specify an option to the
fencing agent. Just because there is no DEFAULT setting doesn't mean that you
can't use this as a fencing agent or that you can't use it in production.
[root@localhost ~]# cat << EOF | fence_ipmilan
failed: operation must be 'on', 'off', or 'reboot'
[root@trin-02 ~]# cat << EOF | fence_ipmilan
ipmilan: ipmitool not found!
failed: Failed to initialize
Unfortunately I don't have access to ipmi hardware at this instant to fully
demonstrate just how one would use this in production, but you should get the
In other words, just add the option that the agent is complaining about missing.
"fence_node mymachine" will give you something like that in the syslog:
Fence of "mymachine" was unsuccessful
fence_node is what cman uses, and doesn't pass options to the underlying fence
method. Without the patch above, fence_node is useless for machines using ipmi lan.
"fence_node mymachine" queries ccs which then gets it's info from cluster.conf
and then passes the information to the agent on stdin. The problem that I am
guessing that you are running into at the moment is that you don't have the
"option" argument specifies in /etc/cluster/cluster.conf. Adding that option to
the cluster.conf will make the agent work for you.
The option would look something like this:
<fencedevice agent="fence_ipmilan" name="ipmi" ipaddr="18.104.22.168"
passwd="password" login="root" action="reboot" />
Adding "action" will make fence_ipmilan work for your setup :)
don't forget to run ccs_update after modifying it. You can also try using
system-config-cluster, but I don't know how that will work.
Fair enough, as long as the bug is properly fixed. Thanks.
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on the solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.