[openstack-dev] blueprint proposal nova-compute fencing for HA ?
GLIKSON at il.ibm.com
Tue Apr 23 10:04:57 UTC 2013
Pacemaker (and the underlying heartbeat mechanism) is a very good tool,
but to the best of my knowledge it is not very scalable. Therefore, it is
a great fit to implement HA of things like mysql, between small number of
'controller' nodes. But when it comes to cloud-scale heartbeat/membership,
there are better tools -- such as ZooKeeper (which is one of the backend
implementations of the service heartbeat mechanism within Nova).
From: Leen Besselink <ubuntu at consolejunkie.net>
To: OpenStack Development Mailing List
<openstack-dev at lists.openstack.org>,
Date: 23/04/2013 10:39 AM
Subject: Re: [openstack-dev] blueprint proposal nova-compute
fencing for HA ?
> I was only talking about the fencing off a compute node part, since
> that's what you started the thread with. :-)
I know I'm going in circles, just trying to get a feel for the best way to
> Presumably you would still use nova APIs that already exist to move the
> instances elsewhere. An 'evacuate' API went in to grizzly for this.
So when any node fails in a Pacemaker cluster, you fence the node, tell
OpenStack about the
failed node and call evacuate for all the instances. The scheduler will
just place them anywhere
(there is already a blueprint for evacuate to call the scheduler and even
an other for handling a whole node)
So, yeah, maybe that is enough.
I guess I was hoping all machines would be the same. Now I'll need to make
clusters. To OpenStack
they will still all look the same I guess.
But it will work with existing tested code, that is also important.
So no blueprint for fencing.
Thank you for the discussion.
OpenStack-dev mailing list
OpenStack-dev at lists.openstack.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the OpenStack-dev