A callback interface about how to fail a job.
A checkpoint ID counter.
A factory for per Job checkpoint recovery components.
A bounded LIFO-queue of
The interface for hooks that can be called by the checkpoint coordinator when triggering or restoring a checkpoint.
A factory to instantiate a
Interface that allows to implement different strategies for repartitioning of operator state as parallelism changes.
Base class for checkpoint statistics.
The checkpoint coordinator coordinates the distributed snapshots of operators and state.
This actor listens to changes in the JobStatus and activates or deactivates the periodic checkpoint scheduler.
The checkpoint failure manager which centralized manage checkpoint failure processing logic.
Encapsulates all the meta data for a checkpoint.
A collection of simple metrics, around the triggering of a checkpoint.
Options for performing the checkpoint.
The configuration of a checkpoint.
A utility class with the methods to write/load/dispose the checkpoint and savepoint metadata.
Counts of checkpoints.
An array based history of checkpoint stats.
A snapshot of the checkpoint stats.
Tracker for checkpoint statistics.
A CompletedCheckpoint describes a checkpoint after all required tasks acknowledged it (with their state) and that is considered successful.
Statistics for a successfully completed checkpoint.
Summary over all completed checkpoints.
Statistics for a failed checkpoint.
This class encapsulates the data from the job manager to restore a task.
Simple encapsulation of state generated by checkpoint coordinator.
Helper for keeping track of min/max/average summaries.
Simple container class which contains the raw/managed operator state and key-group state handles from all sub tasks of an operator and therefore represents the complete state of a logical operator.
This class encapsulates the state for one parallel instance of an operator.
A pending checkpoint is a checkpoint that has been started, but has not been acknowledged by all tasks that need to acknowledge it.
Statistics for a pending checkpoint that is still in progress.
This class is a wrapper over multiple alternative
Statistics for a restored checkpoint.
Current default implementation of
This class encapsulates the operation of assigning restored state when restoring from a checkpoint.
|StateObjectCollection<T extends StateObject>||
This class represents a generic collection for
Container for the chained state of one parallel subtask of an operator/task.
Statistics for a single subtask that is part of a checkpoint.
Internal class for savepoint backwards compatibility.
This class encapsulates state handles to the snapshots of all operator instances executed within one task.
Statistics for a single task/operator that gathers all statistics of its subtasks and provides summary statistics about all subtasks.
Summary of the subtask stats of a single task/operator.
Various reasons why a checkpoint was failure.
Policy for whether checkpoints are retained after a job terminates.
Status of the tracked checkpoint.
The type of checkpoint to perform.
Result of the
Base class for checkpoint related exceptions.
Copyright © 2014–2019 The Apache Software Foundation. All rights reserved.