caosurong created FLINK-40143:
---------------------------------
Summary: Dashboard shows incorrect restart strategy for jobs
configured with restart-strategy options
Key: FLINK-40143
URL: https://issues.apache.org/jira/browse/FLINK-40143
Project: Flink
Issue Type: Bug
Components: Runtime / Configuration, Runtime / REST, Runtime / Web
Frontend
Affects Versions: 1.20.5, 1.19.1
Environment: Reproduced locally with Apache Flink 1.20.5 and 1.19.1
using a MiniCluster and the real Dashboard Job Configuration page. The same
issue was not reproducible in Apache Flink 1.18.1.
Evidence screenshots are available for 1.20.5 and 1.19.1.
Reporter: caosurong
Attachments: image-2026-07-14-11-29-56-872.png,
image-2026-07-14-11-30-18-947.png
When a job is configured with restart-strategy.type=fixed-delay and related
restart-strategy.fixed-delay.* options, the runtime uses the job-level restart
strategy correctly. However, the Dashboard Job Configuration page and the
/jobs/<jid>/config REST
!image-2026-07-14-11-30-18-947.png! response display "Cluster level default
restart strategy".
This is reproducible in Flink 1.19.1 and Flink 1.20.5 LTS. I also checked Flink
1.18.1, where the same issue is not reproducible.
Example job-level configuration:
- restart-strategy.type = fixed-delay
- restart-strategy.fixed-delay.attempts = 10
- restart-strategy.fixed-delay.delay = 1 min
Expected:
The Job Configuration page and the /jobs/<jid>/config REST response should
display the configured fixed-delay restart strategy.
Actual:
The Job Configuration page and the /jobs/<jid>/config REST response display
"Cluster level default restart strategy".
This looks like a regression introduced by the restart strategy configuration
migration in Flink 1.19: runtime scheduling reads RestartStrategyOptions from
Configuration correctly, but ArchivedExecutionConfig still derives the
displayed restart strategy from ExecutionConfig#getRestartStrategy(), which
returns the fallback restart strategy.
I searched existing FLINK issues with keywords including "Cluster level default
restart strategy", "ArchivedExecutionConfig", "restart-strategy Dashboard",
"Job Configuration restart strategy", and "RestartStrategyOptions
ArchivedExecutionConfig", but did not find an existing issue for this Dashboard
display problem.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
