Ask Your Question

Gorg's profile - activity

2016-03-23 01:34:26 -0500 received badge  Popular Question (source)
2015-05-24 04:26:15 -0500 received badge  Popular Question (source)
2015-05-24 04:26:15 -0500 received badge  Notable Question (source)
2015-05-24 04:26:15 -0500 received badge  Famous Question (source)
2015-05-22 12:41:30 -0500 commented question How can I deploy multi-node lab with DevStack with two separate interfaces for each worker node

By flat dhcp network, I mean the network from which IPs are assigned to each VM in order to be externally accessible (e.g via ssh). I have no problem installing Horizon in the controller, the problem is that by following the guide for multi-node Devstack, Horizon is also installed on compute nodes.

2015-05-22 11:58:26 -0500 asked a question How can I deploy multi-node lab with DevStack with two separate interfaces for each worker node

Hi,

I am interested in building a lab with DevStack. I am going to use one PC as cloud controller (one interface on the management network) and two compute nodes (each one with two interfaces, one attacked on the management net and the other one on the flat DHCP net). You can find here one picture with the topology that I would like to build: https://dl.dropboxusercontent.com/u/5366560/flat.jpg (https://dl.dropboxusercontent.com/u/5...). I also want to go with the nova-network option. The think is that the multi-node guide from http://devstack.org/guides/multinode-lab.html (http://devstack.org/guides/multinode-...) does not explain how to do it using two network interfaces on each compute node (one for the management and the other one for the flat DHCP network).

Also when I used the aforementioned guide the controller node was used as a compute node as well (meaning an instance could run on my controller) which is something that I do not want. Furthermore by following the same guide the horizon service was installed on my compute nodes which is something that I don't want as well.

Do you have any ideas about how can I use DevStack: a)to consider two interfaces on each compute node? b)to disable horizon from compute nodes and disable nova-compute service from controller (in order for instances to be launched only on compute nodes)

Thanks a lot

2015-05-20 10:53:35 -0500 received badge  Enthusiast
2015-04-14 16:53:18 -0500 asked a question Develop a security driven scheduler for OpenStack Compute (nova)

Hi,

My goal is to develop a new scheduler that would be able to allocate instances on a physical machine based on the tenants of the virtual machines that are currently running on a physical node. The general idea is that each tenant has some adversaries with whom does NOT want to share a physical machine. For example if companies A and B are competitors, my scheduler should never instantiate instances of A and B on the same physical node. For example when B demands the instantiation of a new instance my scheduler should exclude all physical nodes where instances of A are running. I am thinking of installing a development environment (DevStack) to work on it and for testing purposes.

Is the development of such scheduler possible on OpenStack? Could you please provide any ideas about how something like this could be done? Any kind of help would be greatly appreciated.