HA with core/scheduler/sunstone on cloud

Hi,
this is a double question.

First one: what’s happen when I have an OpenNebula installation and the host with core/scheduler services goes down? I am interested to know if VM in the cluster continues to work.

Second one: imagine I have a cloud hosting that offers me HA services. So I install the opennebula core/scheduler on a VM of that cloud hosting provider and I have solved the HA problem for the manager.
But can manager be put far away from server it manages?

Thanks,
Mario

I have seen no replies so I suppose I was not clear.

I want to know:

  1. what happens to vm if oned goes down
  2. if I can put the server with oned “far away” from servers with virtual machines.

Oned manages the cluster nodes, virtual networks, virtual machines, users, groups and storage datastores. So presumably if it goes there is nothing to control the virtual machines. One way to find out is to test your question. Stop Oned and see what happens.

imagine I have a cloud hosting that offers me HA services. So I install
the opennebula core/scheduler on a VM of that cloud hosting provider and
I have solved the HA problem for the manager.

I wouldn’t consider that solving any high availability problem. You don’t want anything “Far away” from the virtual machines because you’ll run into latency issues to say the least. Ideally it should all be in the same rack on the same network.

If you want to setup a cloud platform, you need to do it correctly and not cut corners. Otherwise there will be no redudnacy and reliability, you’ll might as well not use OpenNebula.

Nothing. The VMs run on top of the hypervisor, if the font-end crashes VMs continue running. OpenNebula will continue monitoring them after it restarts.

You can, but you will have to plan the infrastructure carefully. For example you don’t want the front-end to access the storage back-ends.