[FLINK-24315] Cannot rebuild watcher thread while the K8S API server is unavailable

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.14.0, 1.13.2
• Fix Version/s: 1.13.3, 1.14.3, 1.15.0
• Component/s: Deployment / Kubernetes
• Labels: pull-request-available
• Assignee: Yangze Guo
• Reporter: ouyangwulin
• Created: 17/Sep/21 06:04
• Updated: 15/Dec/21 01:44
• Resolved: 23/Sep/21 13:28

Description

In native k8s integration, Flink will try to rebuild the watcher thread if the API server is temporarily unavailable. However, if the jitter is longer than the web socket timeout, the rebuilding of the watcher will timeout and Flink cannot handle the pod event correctly.

Issue Links

• links to: GitHub Pull Request #17321
• links to: GitHub Pull Request #17336
• links to: GitHub Pull Request #17337