Comment 3 for bug 1315224

mrda | Should we simply power off nodes post registration if they're on already?

devananda | https://github.com/openstack/ironic/blob/master/ironic/api/controllers/v1/node.py#L637 and
devananda | https://github.com/openstack/ironic/blob/master/ironic/conductor/manager.py#L97
devananda | so... I think it's a reasonable thing to do IFF force_power_state_during_sync=TRUE
devananda | even if that option is TRUE, the creation of a node is simply a DB insert. it doesn't pass anything to conductor today -- so this would be a bit of a larger change

mrda | ok, so when a node is registered, if force_power_state_during_sync is True, power off the node (since the default power state in the DB is off)?

devananda | and create another point where API requests could starve RPC threads

mrda | ok, so it sounds like you're reluctant for this change to happen
mrda | ...unless some clever code can be constructed to not create a new problem :)

devananda | https://github.com/openstack/ironic/blob/master/ironic/conductor/manager.py#L530
devananda | so current behavior is, if node has no power state recorded in the db, we just update the DB with the actual state regardless
devananda | so this is a behavior change, and possibly an RPC API change