Comment 1 for bug 1706772

Actually this wasn't caused by the migrate instance flavors change mentioned above, it was happening before that as well:

http://logs.openstack.org/80/461480/5/check/gate-tempest-dsvm-ironic-ipa-wholedisk-bios-agent_ipmitool-tinyipa-ubuntu-xenial-nv/834477a/logs/screen-n-cpu.txt.gz?level=TRACE#_Jul_19_03_34_01_687299

So something has thrown off the timing of when nova-compute starts up and when the ironic-api is available.

This wasn't an issue in Ocata. From Ocata CI logs, the init_host in the driver happens around here:

http://logs.openstack.org/63/485263/2/check/gate-tempest-dsvm-ironic-ipa-wholedisk-bios-agent_ipmitool-tinyipa-ubuntu-xenial-nv/de2c924/logs/screen-n-cpu.txt.gz#_2017-07-21_12_07_20_885

And then get_available_nodes is here:

http://logs.openstack.org/63/485263/2/check/gate-tempest-dsvm-ironic-ipa-wholedisk-bios-agent_ipmitool-tinyipa-ubuntu-xenial-nv/de2c924/logs/screen-n-cpu.txt.gz#_2017-07-21_12_07_21_167

So that's not blowing up like init_host is blowing up in Pike.