[Pacemaker] Enable remote monitoring

Gao,Yan ygao at suse.com
Sun Dec 9 04:05:38 UTC 2012


On 12/07/12 23:47, David Vossel wrote:
> 
> 
> ----- Original Message -----
>> From: "Lars Marowsky-Bree" <lmb at suse.com>
>> To: "The Pacemaker cluster resource manager" <pacemaker at oss.clusterlabs.org>
>> Sent: Friday, December 7, 2012 5:23:07 AM
>> Subject: Re: [Pacemaker] Enable remote monitoring
>>
>> On 2012-12-07T10:38:44, Andrew Beekhof <andrew at beekhof.net> wrote:
>>
>>>> Uhm. Would "container" imply ordering + colocation, or would we
>>>> still
>>>> need them grouped (resource_set'ed, whatever)?
>>> Ordering: absolutely
>>> Colocation is less clear, I think the default is no but David has
>>> suggested an additional meta attribute to turn it on.
>>
>> blackbox - colocated on the node where the container is running
>>            - not probed
> 
> this makes sense to me.
> 
>> whitebox - colocated on the virtual node provided by the container
>>            - probed only there
> 
> I'm not sure about this one though. The whitebox case isn't a special case really in regards to probes.  If we see this resource is running somewhere else... well then it actually is.  We should be able to get away with just having the colocation for the whitebox.
Since a nagios resource is actually a monitor itself, it seems that we
don't have to probe it. Of course if needed, we can introduce a "probe"
meta attribute for primitive, and:

black implies - ordering, colocated, probe="false"
white implies - ordering, probe="false"

Regards,
  Gao,Yan
-- 
Gao,Yan <ygao at suse.com>
Software Engineer
China Server Team, SUSE.




More information about the Pacemaker mailing list