Roland de Boo created AIRFLOW-5312:
--------------------------------------
Summary: KubernetesPodOperator hangs when kubernetes API times out
Key: AIRFLOW-5312
URL: https://issues.apache.org/jira/browse/AIRFLOW-5312
Project: Apache Airflow
Issue Type: Bug
Components: contrib
Affects Versions: 1.10.4
Reporter: Roland de Boo
The KubernetesPodOperator launches a Pod in Kubernetes and then continues to
watch its state and stream its logs. This is done via the Kubernetes Api. Not
all calls to the Kubernetes Api have a timeout defined, so the operator will
wait indefinitely for a response and therefore get stuck.
This is the same issue as https://issues.apache.org/jira/browse/AIRFLOW-5282
but now in the KubernetesPodOperator.
The solution is to add a timeout on all Api calls, and handle the errors
(catching them, adding tenacity retrying where appropriate).
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
