[Pacemaker] Fixed! - Re: Problem with dual-PDU fencing node with redundant PSUs

Vladislav Bogdanov bubble at hoster-ok.com
Tue Jul 2 05:30:18 UTC 2013


02.07.2013 03:10, Andrew Beekhof wrote:
> 
> On 02/07/2013, at 8:51 AM, Andrew Beekhof <andrew at beekhof.net> wrote:
> 
>>
>> On 01/07/2013, at 10:19 PM, Vladislav Bogdanov <bubble at hoster-ok.com> wrote:
>>
>>> 01.07.2013 15:10, Andrew Beekhof wrote:
>>>
>>>>
>>>> And if people start using it, then we might look at simplifying it.
>>>
>>> May be it's worth to have anonymous poll at clusterlabs.org for that?
>>
>> I'll try and put one up today
> 
> http://clusterlabs.org/polls/

IMHO poll question there differs from what we discuss here.
Probably it is very hard to transform this issue into just one question
without loosing some important details.

For me, example of redundant fencing devices are IPMI and PDU. Use one
OR other = redundancy. This is already supported with fencing_topology.

PSU/PDU/circuit pairs are different from that. Use one AND another. I
would not say it is "redundant fencing device". Probably you can replace
that poll question with "How important is support for multi-<something>
fencing devices (like redundant PDUs)". Can't find adequate term tough.






More information about the Pacemaker mailing list