[Pacemaker] Trouble getting fence_apc working on RHEL 6.1

[Digimer]

On 07/18/2011 08:02 PM, Andrew Beekhof wrote:
> On Tue, Jul 19, 2011 at 9:42 AM, Digimer <linux at alteeve.com> wrote:
>> On 07/18/2011 07:26 PM, Andrew Beekhof wrote:
>>> On Mon, Jul 18, 2011 at 7:07 AM, Uwe Grawert <grawert at b1-systems.de> wrote:
>>>> Am 17.07.11 22:49, schrieb Digimer:
>>>>> On 07/17/2011 04:36 PM, Uwe Grawert wrote:
>>>>>> Am 17.07.11 20:32, schrieb Digimer:
>>>>>>> I've been trying to get my APC switched PDU working with Pacemaker using
>>>>>>> Red Hat's fence_apc fence agent. The agent itself works fine. I can
>>>>>>> fence nodes using:
>>>>>>>
>>>>>>> fence_na -a 192.168.1.6 -l apc -p secret -n 1 -o reboot
>>>>>>>
>>>>>>> When I try to fence using CRM though, I get these two errors:
>>>>>>>
>>>>>>> st_output="Parse error: Ignoring unknown option
>>>>>>> 'nodename=an-node02.alteeve.com'#012Failed: Unable to obtain correct
>>>>>>> plug status or plug is not available#012"
>>>>>>>
>>>>>>> The configuration is:
>>>>>>>
>>>>>>> # crm configure show
>>>>>>> node an-node01.alteeve.com \
>>>>>>>     attributes standby="off"
>>>>>>> node an-node02.alteeve.com \
>>>>>>>     attributes standby="off"
>>>>>>> primitive apc-fencing stonith:fence_apc \
>>>>>>>     params \
>>>>>>>     pcmk_host_map="an-node01.alteeve.com:1,an-node02.alteeve.com:2" \
>>>>>>>     pcmk_host_list="an-node01.alteeve.com an-node02.alteeve.com" \
>>>>>>>     pcmk_host_check="static-list" \
>>>>>>>     ipaddr="192.168.1.6" \
>>>>>>>     action="reboot" \
>>>>>>>     login="apc" \
>>>>>>>     passwd="secret" \
>>>>>>>     port="TBA" \
>>>>>>>     property $id="cib-bootstrap-options" \
>>>>>>>     dc-version="1.1.5-5.el6-01e86afaaa6d4a8c4836f68df80ababd6ca3902f" \
>>>>>>>     cluster-infrastructure="openais" \
>>>>>>>     expected-quorum-votes="2" \
>>>>>>>     stonith-enabled="true"
>>>>>>>
>>>>>>> The full log file and the 'pe-warn' file are quite long and would wrap
>>>>>>> badly in email. So I've put them up here:
>>>>>>>
>>>>>>> http://wiki.alteeve.com/index.php/CRM_troubles
>>>>>>>
>>>>>>> Any idea what magic I am missing? :)
>>>>>>
>>>>>> Pacemaker is adding following parameters: option, nodename, port,
>>>>>> additionally to the ones you have configured. The nodename parameter is
>>>>>> not the problem, it is ignored. But you have a port parameter defined:
>>>>>> port="TBA". That one is getting overwritten with the node's name to
>>>>>> fence. I had the same issue with fence_bladecenter agent and had to
>>>>>> patch it.
>>>>>>
>>>>>> I wrote an email to this list a month ago. Subject: "Pacemaker
>>>>>> additional parameters on stonith plugins"
>>>>>>
>>>>>> But there was no big fuzz ;)
>>>>>
>>>>> As I understood it, the:
>>>>>
>>>>> pcmk_host_map="an-node01.alteeve.com:1,an-node02.alteeve.com:2"
>>>>>
>>>>> was supposed to provide the name -> port mapping when port="TBA" is
>>>>> used... Did you get a config working?
>>>>
>>>> I used a different fencing agent. But the problem is, that your
>>>> configuration of the port parameter port="TBA" will be overwritten and
>>>> endup looking like this: port="an-node02.alteeve.com". And that will
>>>> break the fence agent.
>>>
>>> Mostly right. The pcmk_host_map argument _should_ mean port is set to an int.
>>> If not its a bug.  port=TBA is only to shut the shell up.
>>
>> Any feedback on what might be wrong with my config? :)
> 
> Configs look fine, but the logs don't go back far enough.
> I need to see the device being added to stonith-ng.

I'll delete it, restart pacemaker, start tailing the logs and then add
it again. I'll put the logs up and reply to this as soon as that is done.

Thanks. :)

-- 
Digimer
E-Mail:              digimer at alteeve.com
Freenode handle:     digimer
Papers and Projects: http://alteeve.com
Node Assassin:       http://nodeassassin.org
"At what point did we forget that the Space Shuttle was, essentially,
a program that strapped human beings to an explosion and tried to stab
through the sky with fire and math?"