nova-compute fails to start if is_power_on raises exception

[aeva black]

During startup of nova-compute, if virtual_power_driver is_power_on() raises an exception, it will prevent nova-compute from starting. This can happen when a compute instance exists in the database but the SSH credentials that vpd uses are wrong.

I think a solution is for vpd to convert exception.PowerVMConnectionFailed into an error and log it.

Here is a trace of this error:

2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/service.py", line 147, in run_server
2013-04-13 06:52:02,213.213 4164 TRACE nova server.start()
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/service.py", line 429, in start
2013-04-13 06:52:02,213.213 4164 TRACE nova self.manager.init_host()
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/compute/manager.py", line 630, in init_host
2013-04-13 06:52:02,213.213 4164 TRACE nova self._init_instance(context, instance)
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/compute/manager.py", line 539, in _init_instance
2013-04-13 06:52:02,213.213 4164 TRACE nova drv_state = self._get_power_state(context, instance)
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/compute/manager.py", line 649, in _get_power_state
2013-04-13 06:52:02,213.213 4164 TRACE nova return self.driver.get_info(instance)["state"]
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/virt/baremetal/driver.py", line 363, in get_info
2013-04-13 06:52:02,213.213 4164 TRACE nova if pm.is_power_on():
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/virt/baremetal/virtual_power_driver.py", line 198, in is_power_on
2013-04-13 06:52:02,213.213 4164 TRACE nova if not self._check_for_node():
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/virt/baremetal/virtual_power_driver.py", line 147, in _check_for_node
2013-04-13 06:52:02,213.213 4164 TRACE nova full_node_list = self._get_full_node_list()
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/virt/baremetal/virtual_power_driver.py", line 141, in _get_full_node_list
2013-04-13 06:52:02,213.213 4164 TRACE nova full_list = self._run_command(cmd)
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/virt/baremetal/virtual_power_driver.py", line 225, in _run_command
2013-04-13 06:52:02,213.213 4164 TRACE nova self._set_connection()
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/virt/baremetal/virtual_power_driver.py", line 136, in _set_connection
2013-04-13 06:52:02,213.213 4164 TRACE nova self._connection = connection.ssh_connect(self.connection_data)
2013-04-13 06:52:02,213.213 4164 TRACE nova File "/opt/stack/venvs/nova/local/lib/python2.7/site-packages/nova/virt/powervm/common.py", line 62, in ssh_connect
2013-04-13 06:52:02,213.213 4164 TRACE nova raise exception.PowerVMConnectionFailed()
2013-04-13 06:52:02,213.213 4164 TRACE nova PowerVMConnectionFailed: Connection to PowerVM manager failed
2013-04-13 06:52:02,213.213 4164 TRACE nova
2013-04-13 06:52:02,648.648 4217 INFO nova.manager [-] Skipping periodic task _periodic_update_dns because its interval is negative
2013-04-13 06:52:02,705.705 4217 INFO nova.virt.driver [-] Loading compute driver 'baremetal.driver.BareMetalDriver'
2013-04-13 06:52:02,754.754 INFO nova.openstack.common.rpc.common [req-235af7f1-4c57-42b8-8cc7-cbdd1327f12a None None] Connected to AMQP server on 127.0.0.1:5672
2013-04-13 06:52:02,839.839 4217 AUDIT nova.service [-] Starting compute node (version 2013.2)
2013-04-13 06:52:03,176.176 ERROR nova.compute.manager [req-e06f7f34-a2df-44f2-b3f2-ac299d4087d6 None None] Instance bmtest found in the hypervisor, but not in the database

[Robert Collins (lifeless)]

Similarly to the root disk size one, this isn't really a tripleo issue : we can deliver a great tripleo experience without it.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/27973

[Joe Gordon (jogo)]

since we are in the process of deprecating and removing nova baremetal and the patch is abandoned, closing this bug.