Package org.apache.flink.runtime.checkpoint

Class CheckpointFailureManager

java.lang.Object

org.apache.flink.runtime.checkpoint.CheckpointFailureManager

public class CheckpointFailureManager
extends Object

The checkpoint failure manager which centralized manage checkpoint failure processing logic.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static interface	CheckpointFailureManager.FailJobCallback	A callback interface about how to fail a job.

Field Summary

Fields

Modifier and Type	Field	Description
static final String	EXCEEDED_CHECKPOINT_TOLERABLE_FAILURE_MESSAGE
static final int	UNLIMITED_TOLERABLE_FAILURE_NUMBER

Constructor Summary

Constructors

Constructor	Description
CheckpointFailureManager(int tolerableCpFailureNumber, CheckpointFailureManager.FailJobCallback failureCallback)

Method Summary

Modifier and Type	Method	Description
void	checkFailureCounter(CheckpointException exception, long checkpointId)
void	handleCheckpointException(PendingCheckpoint pendingCheckpoint, CheckpointProperties checkpointProperties, CheckpointException exception, ExecutionAttemptID executionAttemptID, org.apache.flink.api.common.JobID job, PendingCheckpointStats pendingCheckpointStats, CheckpointStatsTracker statsTracker)	Failures on JM: all checkpoints - go against failure counter.
void	handleCheckpointSuccess(long checkpointId)	Handle checkpoint success.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

UNLIMITED_TOLERABLE_FAILURE_NUMBER

public static final int UNLIMITED_TOLERABLE_FAILURE_NUMBER

See Also:

• Constant Field Values

EXCEEDED_CHECKPOINT_TOLERABLE_FAILURE_MESSAGE

public static final String EXCEEDED_CHECKPOINT_TOLERABLE_FAILURE_MESSAGE

See Also:

• Constant Field Values

Constructor Details

CheckpointFailureManager

public CheckpointFailureManager(int tolerableCpFailureNumber,
 CheckpointFailureManager.FailJobCallback failureCallback)

Method Details

handleCheckpointException

public void handleCheckpointException(@Nullable
 PendingCheckpoint pendingCheckpoint,
 CheckpointProperties checkpointProperties,
 CheckpointException exception,
 @Nullable
 ExecutionAttemptID executionAttemptID,
 org.apache.flink.api.common.JobID job,
 @Nullable
 PendingCheckpointStats pendingCheckpointStats,
 CheckpointStatsTracker statsTracker)

Failures on JM:

• all checkpoints - go against failure counter.
• any savepoints - don’t do anything, manual action, the failover will not help anyway.

Failures on TM:

• all checkpoints - go against failure counter (failover might help and we want to notify users).
• sync savepoints - we must always fail, otherwise we risk deadlock when the job cancelation waiting for finishing savepoint which never happens.
• non sync savepoints - go against failure counter (failover might help solve the problem).

Parameters:

pendingCheckpoint - the failed checkpoint if it was initialized already.

checkpointProperties - the checkpoint properties in order to determinate which handle strategy can be used.

exception - the checkpoint exception.

executionAttemptID - the execution attempt id, as a safe guard.

job - the JobID.

pendingCheckpointStats - the pending checkpoint statistics.

statsTracker - the tracker for checkpoint statistics.

checkFailureCounter

public void checkFailureCounter(CheckpointException exception,
 long checkpointId)

handleCheckpointSuccess

public void handleCheckpointSuccess(long checkpointId)

Handle checkpoint success.

Parameters:

checkpointId - the failed checkpoint id used to count the continuous failure number based on checkpoint id sequence.