Gyula Fora created FLINK-26135:
----------------------------------
Summary: Separate job and deployment errors in FlinkDeployment
status
Key: FLINK-26135
URL: https://issues.apache.org/jira/browse/FLINK-26135
Project: Flink
Issue Type: Sub-task
Components: Deployment / Kubernetes
Reporter: Gyula Fora
Assignee: Gyula Fora
At the moment the controller does not validate or tolerate any deployment
errors such as incorrect configurations etc. Those will lead to an exception
loop in the reconcile logic.
There are cases where the job deployment cannot be executed due to incorrect
configuration or other causes. In these cases the job can still be running
correctly so the job status should be OK but we should signal a deployment
error to the user that requires action.
There should be a shared validation logic between the controller and the
webhook that should be applied whenever a new FlinkDeployment update is
received by the controller. If an error is detected in the controller, set the
deployment status to error with a useful message and leave the current job
running.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
