yahoo-eng-team team mailing list archive

[Devananda <1120198@xxxxxxxxxxxxxxxxxx>]

Just had a chat with Vish. Seems that this behavior only occurs when the
compute_filter is disabled (which is true in my test environment). I am
marking this as not-a-bug.

Even though dead compute nodes should be getting filtered out by the
compute filter,  the scheduler also shouldn't be caching them for ever.


** Changed in: nova
       Status: In Progress => Invalid

-- 
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/1120198

Title:
  scheduler may pick deleted baremetal nodes

Status in OpenStack Compute (Nova):
  Invalid

Bug description:
  When deleting a baremetal node, the scheduler may continue to attempt
  to boot instances on that node. The compute manager will refuse these
  requests since the selected node is no longer valid. I have not found
  any mechanism to make the scheduler stop attempting to use the non-
  existent node.

  I suspect the cause of this is that
  ComputeManager.update_available_resource does not actively remove the
  resource because it is no longer returned from
  driver.get_available_nodes, but IMBW....

  
  The error looks like this:

  2013-02-09 09:00:03.240 TRACE nova.openstack.common.rpc.amqp   File "/opt/stack/nova/nova/compute/manager.py", line 700, in _run_instance
  2013-02-09 09:00:03.240 TRACE nova.openstack.common.rpc.amqp     rt = self._get_resource_tracker(node)
  2013-02-09 09:00:03.240 TRACE nova.openstack.common.rpc.amqp   File "/opt/stack/nova/nova/compute/manager.py", line 346, in _get_resource_tracker
  2013-02-09 09:00:03.240 TRACE nova.openstack.common.rpc.amqp     raise exception.NovaException(msg)
  2013-02-09 09:00:03.240 TRACE nova.openstack.common.rpc.amqp NovaException: 2 is not a valid node managed by this compute host.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1120198/+subscriptions