Andrey Volkov 4164518502 Ensure pod logs are fetched in case of exception in any operator ... This patch tries to cover some edge cases could happen during Shipyard Airflow operator execution. All operators at the moment make interactions with other services i.e. k8s pods. In a case of exceptions during execution of the operator, logs will be fetched from the appropriate pod and if the operator has "fetch_failure_details" method (see DrydockBaseOperator) it will be called as well. What exception could happen during an operator execution? Besides explicitly defined in code like DrydockClientUseFailureException, other exception e.g. KeyError or similar may be raised. It's not clear who is a culprit in that client side (Shipyard) or server side (Drydock, Armada, Deckhand, Promenade). So this patch applies defensive mode and gets logs from pods and gets additional details for any exceptional situations. For doing that do_execute method is wrapped with try..except in UcpBaseOperator.execute. While fetching logs from a pod and fetching failure details it makes appropriate logging by itself and finally reraises the original exception. Change-Id: If1501e9a24b05edb6eb32c7b1b2d27f24f3ee063		2018-09-19 09:17:13 -07:00
..
unit	Ensure pod logs are fetched in case of exception in any operator	2018-09-19 09:17:13 -07:00
__init__.py	Refactor shipyard to UCP target layout	2018-04-24 16:47:13 -05:00