Fail safe wsgi logging

[Arathi]

If there are any uncaught exceptions from api controllers, they are handled in nova/api/openstack/__init__.py:FaultWrapper, logged the handled exception and HTTP 500 response is sent to the client.

There could be error while logging these uncaught exceptions and we missout them in the nova logs. In such a case, these exceptions should be logged into syslog for further investigation.

[Arathi (arathi-darshanam)]

For reference changes are at: https://review.openstack.org/#/c/13567/

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/60892

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/61180

[Christopher Yeoh (cyeoh-0)]

The first thing we do is log the exception. This happens before we attempt to return any information to the client or do any other processing. So I don't think we need any other fallback. Unless you have a testcase which demonstrates this is necessary.

[Tom Fifield (fifieldt)]

Cinder patch was abandoned