Jump to: navigation, search


Sometimes hardware just doesn't work as we hope. Rather than letting our users determine it for us, we want to add the ability to add hardware to the system while not adding to the general pool. System users should be able to launch vms on the staged hardware... http://etherpad.openstack.org/CloudHealth

existing system background

Currently when a new compute / volume node shows up, it adds itself to the services table (with the current time in the "last reported at" field and enabled set to true).


1) add workers as "default disabled"

By defaulting the enabled field to false and adding nova-manage commands for toggling enabled. New resources aren't added to the pool for general use.

2) allow system users to specify a specific resource to use to launch (skipping scheduling)

By overriding availability-zone parameter of the aws api for CreateVolume / RunInstance, we can allow admin users to launch instances / volumes on specific nodes.

Propose that the aws api we use "staging.node_name" as the zone.


euca-run-instance ami-tiny -z staging.cleo05

Would launch launch image ami-tiny on node cleo05.