Search before asking
Fluss version
0.9.0 (latest release)
Please describe the bug 🐞
The Fluss Lake Tiering job does not rely on Flink checkpoints so checkpointing is disabled by default.
However, when checkpointing is disabled and no restart strategy is explicitly configured, Flink uses the no-restart strategy by default.
Currently, FlussLakeTiering configures the full-restart failover strategy but does not configure a job restart strategy. When a transient exception occurs during tiering, the Flink job transitions directly to the FAILED state and is not restarted. As a result, the tiering service stops permanently and requires manual job resubmission.
The existing failover integration test enables checkpointing and therefore does not reproduce the actual production configuration where checkpointing is disabled, leaving this issue uncovered.
Solution
No response
Are you willing to submit a PR?
Search before asking
Fluss version
0.9.0 (latest release)
Please describe the bug 🐞
The Fluss Lake Tiering job does not rely on Flink checkpoints so checkpointing is disabled by default.
However, when checkpointing is disabled and no restart strategy is explicitly configured, Flink uses the no-restart strategy by default.
Currently, FlussLakeTiering configures the full-restart failover strategy but does not configure a job restart strategy. When a transient exception occurs during tiering, the Flink job transitions directly to the FAILED state and is not restarted. As a result, the tiering service stops permanently and requires manual job resubmission.
The existing failover integration test enables checkpointing and therefore does not reproduce the actual production configuration where checkpointing is disabled, leaving this issue uncovered.
Solution
No response
Are you willing to submit a PR?