Comment 6 for bug 1515870

Right before the scheduler failure in comment 5 I see this in the n-cpu logs:

http://logs.openstack.org/20/407220/2/check/gate-tempest-dsvm-ironic-ipa-wholedisk-agent_ipmitool-tinyipa-multinode-ubuntu-xenial-nv/0779f94/logs/screen-n-cpu.txt.gz#_2017-01-04_19_03_55_573

2017-01-04 19:03:55.573 28458 INFO nova.compute.manager [req-a0b07a5e-2c5a-4550-a0b5-6d1fe2b1d9e0 - -] Deleting orphan compute node 1 hypervisor host is 00d62acd-6f3b-4cb8-9668-12517c84b3b9, nodes are set([u'74396195-3bd0-4bc3-9f51-2f9d9ff32616', u'dd8dcc23-04f4-4718-a079-e1b77d70bad9', u'36827858-740c-4f1d-a0e9-17339f441338', u'86939305-5678-4796-a250-76295faf1890', u'0f813337-25e3-4c12-a2b0-cf2222f3cc7f', u'21da4933-a128-45f6-a765-7e6bc071e0f3'])