Open Stack

Dear team,

At  the university of Zürich switzerland, we are using the Openstack fake  compute driver in the framework of a Master study project on fairness  algorithms in the nova compute.

Our setup: Openstack icehouse: One controller node + 2 compute nodes. These three nodes are virtual using the qemu hypervisor.
In /etc/nova/nova.conf we have specified:
compute_driver = fake.FakeDriver
connection_type=fake

We create vms using names such as fake01_f1, fake02_f2 where f1= default flavor with id=1 f2 = default flavor with id=2

We  are using a dictionary (eg: dic_load={ram=100MB, vcpu=1, disk=1GB}) to  simulate the current load on a vm. Based on this fake load (dic_load as  input) we can easily used the flavors to compute the remaining free  resources on the fake vm. We created a fake_vm_load method for that  purpose.

Question: How and where do we integrate the fake_vm_load  method in openstack to make our simulation dynamic at the cloud level?  Our intention is to be able to call this method on a running fake vm and  pass it a load that will persist as long as the fake vm is running. 

For  example upon typing the command "nova host-describe compute-host1" we  would like the fake loads on all running fake vms of compute-host1 to be  reflected in the output of this command.

Thanks in advance for your assistance.

Regards,

Louis-Marie LOE

  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20140728/12b966ab/attachment.html>