good day all,
we are going to reprovision one of our HPC clusters to openstack, i was
looking at the ubuntu cloud deployment docs
(http://cloud.ubuntu.com/2011/10/ubuntu-cloud-deployment-with-orchestra-and-juju/
and https://help.ubuntu.com/community/UbuntuCloudInfrastructure) and it
indicates inside there that the recommended build is for all nodes
(admin and compute) to be on two physical networks. our cluster is
typical HPC, two admin nodes connected to the WAN/internet and connected
to private cluster networks (both gigE and IB). there are references to
an 'original document' that describes the 'hacking and workarounds'
needed to build on a single network, but i cannot seem to find that
document:
https://help.ubuntu.com/community/UbuntuCloudInfrastructure
Requirements
It should be assumed that the user deploying this setup has all nodes
with two network interfaces linked to two seperate physical networks.
Deploying with a single network is possible with some hacking and
workarounds, see original document for that.
It's also assumed the user has sufficient number of nodes available in
Orchestra. The minimum deployment outlined below requires a minimum of 6
machines including the juju bootstrap node.
looking for any suggestions about deploying in our environment. thanks!
--- michael
--
Ubuntu-cloud mailing list
[email protected]
Modify settings or unsubscribe at:
https://lists.ubuntu.com/mailman/listinfo/ubuntu-cloud
