Home / Flink / Checkpoints and state
Flink: JobException: Recovery is suppressed by NoRestartBackoffTimeStrategy
org.apache.flink.runtime.JobExceptionRecovery is suppressed by NoRestartBackoffTimeStrategyA task failed and the job was not restarted because the restart strategy does not allow it. Without checkpointing, the default restart strategy is "disable", so the first failure ends the job. The cause is the exception attached to this one.
Common causes
- Checkpointing is not enabled, so no restart strategy applies by default.How to confirm: The job configuration has no execution.checkpointing.interval and no restart-strategy.type.
- A restart strategy is set but its attempts are used up (for example fixed-delay with a small attempts value).How to confirm: The message names FixedDelayRestartBackoffTimeStrategy or ExponentialDelayRestartBackoffTimeStrategy instead.
How to fix it
Fix the underlying failure shown in the "Caused by" section first.
Enable checkpointing (which turns on exponential-delay restarts by default) or set a restart strategy explicitly.
| Setting | Default | Note |
|---|---|---|
execution.checkpointing.interval | none | |
restart-strategy.type | none | disable without checkpointing, exponential-delay with it |
Where the message comes from
ExecutionFailureHandler.javainapache/flinkatrelease-2.3.0- Message checked in Flink 2.3.0
Message template in the source: Recovery is suppressed by %s