[Pacemaker] Andrew and Lars please confirm this.

Andrew Beekhof beekhof at gmail.com
Wed Mar 4 09:47:29 UTC 2009


On Wed, Mar 4, 2009 at 10:30, Romi Verma <romi3rdfeb at gmail.com> wrote:
>> How does a node in Australia connect to a stonith device in Germany if
>> the network is down?
>> Or more generally, how can the nodes in Australia ensure that the
>> nodes in Germany are not running the same services?
>>
>> How do you even know that the nodes in Australia should take over?
>
> ok so it seems i am missing something here.  lets take an example of two
> nodes cluster. my understanding is , pacemaker cluster will be using one
> network (bind nw interface)  for heartbeat  and if it fails then it , both
> node will be in split brain situation and both will try to fence each
> other.  now for fencing if we are using ilo then , it will be using entirely
> different network

Unless you work for the military, this is clearly not the case if you
have multiple sites at geographically different locations.

Multiple NICS != multiple _independent_ networks.

> and this way it shoud work .  if ilo network becomes down
> , then off course the stonith will not work.  and this will be second
> failure. here our goal is to prevent  single point of failure.
>
> so the above thing should be independent of distance, in spite of we are
> having local cluster , or  extended cluster (whose nodes are spanning across
> multiple sites) , the stonith behavior should be same. am i wrong??

Yes, you are wrong.  Because once you start going over the internet
the "stonith network" and "heartbeat network" share a common point of
failure.




More information about the Pacemaker mailing list