[FLINK-9207] Client returns SUCCESS(0) return code for canceled job

• Type: Bug
• Status: Open
• Priority: Not a Priority
• Resolution: Unresolved
• Affects Version/s: 1.5.0
• Component/s: Command Line Client
• Labels: auto-deprioritized-major auto-deprioritized-minor

Environment

Version: 1.5.0, Commit : 2af481a

• Assignee: Unassigned
• Reporter: Amit Jain
• Created: 18/Apr/18 12:57
• Updated: 08/Dec/21 10:39

Description

Flink Client returns zero return code when a job is deliberately canceled. 

Steps to reproduced it:

1. bin/flink run -p 10 -m yarn-cluster -yjm 1024 -ytm 12288 WordCount.jar

2. User externally canceled the job.

3. Job Manager marked the job as CANCELED.

4. Although client code emits following logs, still returns zero return code.

org.apache.hadoop.yarn.client.api.impl.YarnClientImpl         - Killed application application_1523726493647_.....

Job scheduler like Airflow would have hard-time detecting whether the submitted job was canceled or not.