yahoo-eng-team team mailing list archive

[Sean Dague <sean@xxxxxxxxx>]

Ok, I've been looking at this block of code, and it's terribly
confusing. The big issue is that the selection of whether or not we
should send an update is mixed up with the send parts.

The solution would be a refactor of send_update to extract to the front
all the change determination logic, so we have an answer of "should we
send an update, and what has changed"

** Changed in: nova
       Status: Invalid => Confirmed

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1204169

Title:
  compute instance.update messages sometimes have the wrong values for
  instance state

Status in OpenStack Compute (Nova):
  In Progress

Bug description:
  Compute instance.update messages that are not triggered by a state
  change (e.g. setting the host in the resource tracker) have default
  (None) values for task_state, old_vm_state and old_ task_state.

  This can make the instance state sequence look wrong to anything
  consuming the messages (e.g stacktach)

   compute.instance.update  None(None) -> Building(none)
   scheduler.run_instance.scheduled 
   compute.instance.update  building(None) ->  building(scheduling)
   compute.instance.create.start 
   compute.instance.update  building(None) ->  building(None)
   compute.instance.update  building(None) ->  building(networking)
   compute.instance.update  building(networking) -> building(block_device_mapping)

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1204169/+subscriptions