org.apache.flink.runtime.state.filesystem

Class FsStateBackend

java.lang.Object

org.apache.flink.runtime.state.AbstractStateBackend

org.apache.flink.runtime.state.filesystem.AbstractFileStateBackend

org.apache.flink.runtime.state.filesystem.FsStateBackend

All Implemented Interfaces:

Serializable, CheckpointStorage, ConfigurableStateBackend, StateBackend

Deprecated.

@Deprecated
 @PublicEvolving
public class FsStateBackend
extends AbstractFileStateBackend
implements ConfigurableStateBackend

IMPORTANT FsStateBackend is deprecated in favor of HashMapStateBackend and FileSystemCheckpointStorage. This change does not affect the runtime characteristics of your Jobs and is simply an API change to help better communicate the ways Flink separates local state storage from fault tolerance. Jobs can be upgraded without loss of state. If configuring your state backend via the StreamExecutionEnvironment please make the following changes.

 		StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
 		env.setStateBackend(new HashMapStateBackend());
 		env.getCheckpointConfig().setCheckpointStorage("hdfs:///checkpoints");
 

If you are configuring your state backend via the config.yaml please make the following changes set your state backend type to "hashmap" state.backend.type: hashmap.

This state backend holds the working state in the memory (JVM heap) of the TaskManagers. The state backend checkpoints state as files to a file system (hence the backend's name).

Each checkpoint individually will store all its files in a subdirectory that includes the checkpoint number, such as hdfs://namenode:port/flink-checkpoints/chk-17/.

State Size Considerations

Working state is kept on the TaskManager heap. If a TaskManager executes multiple tasks concurrently (if the TaskManager has multiple slots, or if slot-sharing is used) then the aggregate state of all tasks needs to fit into that TaskManager's memory.

This state backend stores small state chunks directly with the metadata, to avoid creating many small files. The threshold for that is configurable. When increasing this threshold, the size of the checkpoint metadata increases. The checkpoint metadata of all retained completed checkpoints needs to fit into the JobManager's heap memory. This is typically not a problem, unless the threshold getMinFileSizeThreshold() is increased significantly.

Persistence Guarantees

Checkpoints from this state backend are as persistent and available as filesystem that is written to. If the file system is a persistent distributed file system, this state backend supports highly available setups. The backend additionally supports savepoints and externalized checkpoints.

Configuration

As for all state backends, this backend can either be configured within the application (by creating the backend with the respective constructor parameters and setting it on the execution environment) or by specifying it in the Flink configuration.

If the state backend was specified in the application, it may pick up additional configuration parameters from the Flink configuration. For example, if the backend if configured in the application without a default savepoint directory, it will pick up a default savepoint directory specified in the Flink configuration of the running job/cluster. That behavior is implemented via the configure(ReadableConfig, ClassLoader) method.

See Also:

Serialized Form

Nested Class Summary

Nested classes/interfaces inherited from interface org.apache.flink.runtime.state.StateBackend

StateBackend.CustomInitializationMetrics, StateBackend.KeyedStateBackendParameters<K>, StateBackend.OperatorStateBackendParameters

Field Summary

Fields inherited from class org.apache.flink.runtime.state.AbstractStateBackend

latencyTrackingConfigBuilder

Constructor Summary

Constructors

Constructor and Description
FsStateBackend(Path checkpointDataUri) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.
FsStateBackend(Path checkpointDataUri, boolean asynchronousSnapshots) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.
FsStateBackend(String checkpointDataUri) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.
FsStateBackend(String checkpointDataUri, boolean asynchronousSnapshots) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.
FsStateBackend(URI checkpointDataUri) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.
FsStateBackend(URI checkpointDataUri, boolean asynchronousSnapshots) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.
FsStateBackend(URI checkpointDataUri, int fileStateSizeThreshold) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.
FsStateBackend(URI checkpointDataUri, int fileStateSizeThreshold, boolean asynchronousSnapshots) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.
FsStateBackend(URI checkpointDataUri, URI defaultSavepointDirectory) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.
FsStateBackend(URI checkpointDirectory, URI defaultSavepointDirectory, int fileStateSizeThreshold, int writeBufferSize, TernaryBoolean asynchronousSnapshots) Deprecated. Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

Method Summary

Modifier and Type	Method and Description
FsStateBackend	configure(ReadableConfig config, ClassLoader classLoader) Deprecated. Creates a copy of this state backend that uses the values defined in the configuration for fields where that were not specified in this state backend.
CheckpointStorageAccess	createCheckpointStorage(JobID jobId) Deprecated. Creates a storage for checkpoints for the given job.
<K> AbstractKeyedStateBackend<K>	createKeyedStateBackend(StateBackend.KeyedStateBackendParameters<K> parameters) Deprecated. Creates a new CheckpointableKeyedStateBackend that is responsible for holding keyed state and checkpointing it.
OperatorStateBackend	createOperatorStateBackend(StateBackend.OperatorStateBackendParameters parameters) Deprecated. Creates a new OperatorStateBackend that can be used for storing operator state.
Path	getBasePath() Deprecated. Deprecated in favor of getCheckpointPath().
Path	getCheckpointPath() Deprecated. Gets the base directory where all the checkpoints are stored.
int	getMinFileSizeThreshold() Deprecated. Gets the threshold below which state is stored as part of the metadata, rather than in files.
int	getWriteBufferSize() Deprecated. Gets the write buffer size for created checkpoint stream.
boolean	isUsingAsynchronousSnapshots() Deprecated. Gets whether the key/value data structures are asynchronously snapshotted, which is always true for this state backend.
boolean	supportsNoClaimRestoreMode() Deprecated. Tells if a state backend supports the RestoreMode.NO_CLAIM mode.
boolean	supportsSavepointFormat(SavepointFormatType formatType) Deprecated.
String	toString() Deprecated.

Methods inherited from class org.apache.flink.runtime.state.filesystem.AbstractFileStateBackend

getSavepointPath, resolveCheckpoint

Methods inherited from class org.apache.flink.runtime.state.AbstractStateBackend

getCompressionDecorator

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.apache.flink.runtime.state.StateBackend

createAsyncKeyedStateBackend, getName, supportsAsyncKeyedStateBackend, useManagedMemory

Constructor Detail

FsStateBackend

public FsStateBackend(String checkpointDataUri)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDataUri - The URI describing the filesystem (scheme and optionally authority), and the path to the checkpoint data directory.

FsStateBackend

public FsStateBackend(String checkpointDataUri,
                      boolean asynchronousSnapshots)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDataUri - The URI describing the filesystem (scheme and optionally authority), and the path to the checkpoint data directory.

asynchronousSnapshots - This parameter is only there for API compatibility. Checkpoints are always asynchronous now.

FsStateBackend

public FsStateBackend(Path checkpointDataUri)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDataUri - The URI describing the filesystem (scheme and optionally authority), and the path to the checkpoint data directory.

FsStateBackend

public FsStateBackend(Path checkpointDataUri,
                      boolean asynchronousSnapshots)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDataUri - The URI describing the filesystem (scheme and optionally authority), and the path to the checkpoint data directory.

asynchronousSnapshots - This parameter is only there for API compatibility. Checkpoints are always asynchronous now.

FsStateBackend

public FsStateBackend(URI checkpointDataUri)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDataUri - The URI describing the filesystem (scheme and optionally authority), and the path to the checkpoint data directory.

FsStateBackend

public FsStateBackend(URI checkpointDataUri,
                      @Nullable
                      URI defaultSavepointDirectory)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI. Optionally, this constructor accepts a default savepoint storage directory to which savepoints are stored when no custom target path is give to the savepoint command.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDataUri - The URI describing the filesystem (scheme and optionally authority), and the path to the checkpoint data directory.

defaultSavepointDirectory - The default directory to store savepoints to. May be null.

FsStateBackend

public FsStateBackend(URI checkpointDataUri,
                      boolean asynchronousSnapshots)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDataUri - The URI describing the filesystem (scheme and optionally authority), and the path to the checkpoint data directory.

asynchronousSnapshots - This parameter is only there for API compatibility. Checkpoints are always asynchronous now.

FsStateBackend

public FsStateBackend(URI checkpointDataUri,
                      int fileStateSizeThreshold)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDataUri - The URI describing the filesystem (scheme and optionally authority), and the path to the checkpoint data directory.

fileStateSizeThreshold - State up to this size will be stored as part of the metadata, rather than in files

FsStateBackend

public FsStateBackend(URI checkpointDataUri,
                      int fileStateSizeThreshold,
                      boolean asynchronousSnapshots)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDataUri - The URI describing the filesystem (scheme and optionally authority), and the path to the checkpoint data directory.

fileStateSizeThreshold - State up to this size will be stored as part of the metadata, rather than in files (-1 for default value).

asynchronousSnapshots - This parameter is only there for API compatibility. Checkpoints are always asynchronous now.

FsStateBackend

public FsStateBackend(URI checkpointDirectory,
                      @Nullable
                      URI defaultSavepointDirectory,
                      int fileStateSizeThreshold,
                      int writeBufferSize,
                      TernaryBoolean asynchronousSnapshots)

Deprecated.

Creates a new state backend that stores its checkpoint data in the file system and location defined by the given URI.

A file system for the file system scheme in the URI (e.g., 'file://', 'hdfs://', or 'S3://') must be accessible via FileSystem.get(URI).

For a state backend targeting HDFS, this means that the URI must either specify the authority (host and port), or that the Hadoop configuration that describes that information must be in the classpath.

Parameters:

checkpointDirectory - The path to write checkpoint metadata to.

defaultSavepointDirectory - The path to write savepoints to. If null, the value from the runtime configuration will be used, or savepoint target locations need to be passed when triggering a savepoint.

fileStateSizeThreshold - State below this size will be stored as part of the metadata, rather than in files. If -1, the value configured in the runtime configuration will be used, or the default value (1KB) if nothing is configured.

writeBufferSize - Write buffer size used to serialize state. If -1, the value configured in the runtime configuration will be used, or the default value (4KB) if nothing is configured.

asynchronousSnapshots - This parameter is only there for API compatibility. Checkpoints are always asynchronous now.

Method Detail

getBasePath

@Deprecated
public Path getBasePath()

Deprecated. Deprecated in favor of getCheckpointPath().

Gets the base directory where all the checkpoints are stored. The job-specific checkpoint directory is created inside this directory.

Returns:

The base directory for checkpoints.

getCheckpointPath

@Nonnull
public Path getCheckpointPath()

Deprecated.

Gets the base directory where all the checkpoints are stored. The job-specific checkpoint directory is created inside this directory.

Overrides:

getCheckpointPath in class AbstractFileStateBackend

Returns:

The base directory for checkpoints.

getMinFileSizeThreshold

public int getMinFileSizeThreshold()

Deprecated.

Gets the threshold below which state is stored as part of the metadata, rather than in files. This threshold ensures that the backend does not create a large amount of very small files, where potentially the file pointers are larger than the state itself.

If not explicitly configured, this is the default value of CheckpointingOptions.FS_SMALL_FILE_THRESHOLD.

Returns:

The file size threshold, in bytes.

getWriteBufferSize

public int getWriteBufferSize()

Deprecated.

Gets the write buffer size for created checkpoint stream.

If not explicitly configured, this is the default value of CheckpointingOptions.FS_WRITE_BUFFER_SIZE.

Returns:

The write buffer size, in bytes.

isUsingAsynchronousSnapshots

public boolean isUsingAsynchronousSnapshots()

Deprecated.

Gets whether the key/value data structures are asynchronously snapshotted, which is always true for this state backend.

supportsNoClaimRestoreMode

public boolean supportsNoClaimRestoreMode()

Deprecated.

Description copied from interface: StateBackend

Tells if a state backend supports the RestoreMode.NO_CLAIM mode.

If a state backend supports NO_CLAIM mode, it should create an independent snapshot when it receives CheckpointType.FULL_CHECKPOINT in Snapshotable.snapshot(long, long, CheckpointStreamFactory, CheckpointOptions).

Specified by:

supportsNoClaimRestoreMode in interface StateBackend

Returns:

If the state backend supports RestoreMode.NO_CLAIM mode.

supportsSavepointFormat

public boolean supportsSavepointFormat(SavepointFormatType formatType)

Deprecated.

Specified by:

supportsSavepointFormat in interface StateBackend

configure

public FsStateBackend configure(ReadableConfig config,
                                ClassLoader classLoader)

Deprecated.

Creates a copy of this state backend that uses the values defined in the configuration for fields where that were not specified in this state backend.

Specified by:

configure in interface ConfigurableStateBackend

Parameters:

config - the configuration

classLoader - The class loader that should be used to load the state backend.

Returns:

The re-configured variant of the state backend

createCheckpointStorage

public CheckpointStorageAccess createCheckpointStorage(JobID jobId)
                                                throws IOException

Deprecated.

Description copied from interface: CheckpointStorage

Creates a storage for checkpoints for the given job. The checkpoint storage is used to write checkpoint data and metadata.

Specified by:

createCheckpointStorage in interface CheckpointStorage

Parameters:

jobId - The job to store checkpoint data for.

Returns:

A checkpoint storage for the given job.

Throws:

IOException - Thrown if the checkpoint storage cannot be initialized.

createKeyedStateBackend

public <K> AbstractKeyedStateBackend<K> createKeyedStateBackend(StateBackend.KeyedStateBackendParameters<K> parameters)
                                                         throws BackendBuildingException

Deprecated.

Description copied from interface: StateBackend

Creates a new CheckpointableKeyedStateBackend that is responsible for holding keyed state and checkpointing it.

Keyed State is state where each value is bound to a key.

Specified by:

createKeyedStateBackend in interface StateBackend

Specified by:

createKeyedStateBackend in class AbstractStateBackend

Type Parameters:

K - The type of the keys by which the state is organized.

Parameters:

parameters - The arguments bundle for creating CheckpointableKeyedStateBackend.

Returns:

The Keyed State Backend for the given job, operator, and key group range.

Throws:

BackendBuildingException

createOperatorStateBackend

public OperatorStateBackend createOperatorStateBackend(StateBackend.OperatorStateBackendParameters parameters)
                                                throws BackendBuildingException

Deprecated.

Description copied from interface: StateBackend

Creates a new OperatorStateBackend that can be used for storing operator state.

Operator state is state that is associated with parallel operator (or function) instances, rather than with keys.

Specified by:

createOperatorStateBackend in interface StateBackend

Specified by:

createOperatorStateBackend in class AbstractStateBackend

Parameters:

parameters - The arguments bundle for creating OperatorStateBackend.

Returns:

The OperatorStateBackend for operator identified by the job and operator identifier.

Throws:

BackendBuildingException

toString

public String toString()

Deprecated.

Overrides:

toString in class Object