Package org.apache.flink.state.rocksdb

Class EmbeddedRocksDBStateBackend

java.lang.Object
org.apache.flink.runtime.state.AbstractStateBackend
org.apache.flink.runtime.state.AbstractManagedMemoryStateBackend
org.apache.flink.state.rocksdb.EmbeddedRocksDBStateBackend
All Implemented Interfaces:
Serializable, org.apache.flink.runtime.state.ConfigurableStateBackend, org.apache.flink.runtime.state.StateBackend
Direct Known Subclasses:
EmbeddedRocksDBStateBackend
@PublicEvolving public class EmbeddedRocksDBStateBackend extends org.apache.flink.runtime.state.AbstractManagedMemoryStateBackend implements org.apache.flink.runtime.state.ConfigurableStateBackend
A StateBackend that stores its state in an embedded RocksDB instance. This state backend can store very large state that exceeds memory and spills to local disk. All key/value state (including windows) is stored in the key/value index of RocksDB. For persistence against loss of machines, please configure a CheckpointStorage instance for the Job.

The behavior of the RocksDB instances can be parametrized by setting RocksDB Options using the methods setPredefinedOptions(PredefinedOptions) and setRocksDBOptions(RocksDBOptionsFactory).

See Also:

  • Constructor Details

    • EmbeddedRocksDBStateBackend

      public EmbeddedRocksDBStateBackend()
      Creates a new EmbeddedRocksDBStateBackend for storing local state.

    • EmbeddedRocksDBStateBackend

      public EmbeddedRocksDBStateBackend(boolean enableIncrementalCheckpointing)
      Creates a new EmbeddedRocksDBStateBackend for storing local state.
      Parameters:
      enableIncrementalCheckpointing - True if incremental checkpointing is enabled.

    • EmbeddedRocksDBStateBackend

      public EmbeddedRocksDBStateBackend(org.apache.flink.util.TernaryBoolean enableIncrementalCheckpointing)
      Creates a new EmbeddedRocksDBStateBackend for storing local state.
      Parameters:
      enableIncrementalCheckpointing - True if incremental checkpointing is enabled.

    • EmbeddedRocksDBStateBackend

      protected EmbeddedRocksDBStateBackend(EmbeddedRocksDBStateBackend original, org.apache.flink.configuration.ReadableConfig config, ClassLoader classLoader)
      Private constructor that creates a re-configured copy of the state backend.
      Parameters:
      original - The state backend to re-configure.
      config - The configuration.
      classLoader - The class loader.

  • Method Details

    • configure

      public EmbeddedRocksDBStateBackend configure(org.apache.flink.configuration.ReadableConfig config, ClassLoader classLoader)
      Creates a copy of this state backend that uses the values defined in the configuration for fields where that were not yet specified in this state backend.
      Specified by:
      configure in interface org.apache.flink.runtime.state.ConfigurableStateBackend
      Parameters:
      config - The configuration.
      classLoader - The class loader.
      Returns:
      The re-configured variant of the state backend

    • supportsNoClaimRestoreMode

      public boolean supportsNoClaimRestoreMode()
      Specified by:
      supportsNoClaimRestoreMode in interface org.apache.flink.runtime.state.StateBackend

    • supportsSavepointFormat

      public boolean supportsSavepointFormat(org.apache.flink.core.execution.SavepointFormatType formatType)
      Specified by:
      supportsSavepointFormat in interface org.apache.flink.runtime.state.StateBackend

    • createKeyedStateBackend

      public <K> org.apache.flink.runtime.state.AbstractKeyedStateBackend<K> createKeyedStateBackend(org.apache.flink.runtime.state.StateBackend.KeyedStateBackendParameters<K> parameters) throws IOException
      Specified by:
      createKeyedStateBackend in interface org.apache.flink.runtime.state.StateBackend
      Specified by:
      createKeyedStateBackend in class org.apache.flink.runtime.state.AbstractManagedMemoryStateBackend
      Throws:
      IOException

    • createOperatorStateBackend

      public org.apache.flink.runtime.state.OperatorStateBackend createOperatorStateBackend(org.apache.flink.runtime.state.StateBackend.OperatorStateBackendParameters parameters) throws Exception
      Specified by:
      createOperatorStateBackend in interface org.apache.flink.runtime.state.StateBackend
      Specified by:
      createOperatorStateBackend in class org.apache.flink.runtime.state.AbstractStateBackend
      Throws:
      Exception

    • getMemoryConfiguration

      public RocksDBMemoryConfiguration getMemoryConfiguration()
      Gets the memory configuration object, which offers settings to control RocksDB's memory usage.

    • setDbStoragePath

      public void setDbStoragePath(String path)
      Sets the path where the RocksDB local database files should be stored on the local file system. Setting this path overrides the default behavior, where the files are stored across the configured temp directories.

      Passing null to this function restores the default behavior, where the configured temp directories will be used.

      Parameters:
      path - The path where the local RocksDB database files are stored.

    • setDbStoragePaths

      public void setDbStoragePaths(String... paths)
      Sets the directories in which the local RocksDB database puts its files (like SST and metadata files). These directories do not need to be persistent, they can be ephemeral, meaning that they are lost on a machine failure, because state in RocksDB is persisted in checkpoints.

      If nothing is configured, these directories default to the TaskManager's local temporary file directories.

      Each distinct state will be stored in one path, but when the state backend creates multiple states, they will store their files on different paths.

      Passing null to this function restores the default behavior, where the configured temp directories will be used.

      Parameters:
      paths - The paths across which the local RocksDB database files will be spread.

    • getDbStoragePaths

      public String[] getDbStoragePaths()
      Gets the configured local DB storage paths, or null, if none were configured.

      Under these directories on the TaskManager, RocksDB stores its SST files and metadata files. These directories do not need to be persistent, they can be ephermeral, meaning that they are lost on a machine failure, because state in RocksDB is persisted in checkpoints.

      If nothing is configured, these directories default to the TaskManager's local temporary file directories.

    • isIncrementalCheckpointsEnabled

      public boolean isIncrementalCheckpointsEnabled()
      Gets whether incremental checkpoints are enabled for this state backend.

    • getPriorityQueueStateType

      public EmbeddedRocksDBStateBackend.PriorityQueueStateType getPriorityQueueStateType()
      Gets the type of the priority queue state. It will fallback to the default value, if it is not explicitly set.
      Returns:
      The type of the priority queue state.

    • setPriorityQueueStateType

      public void setPriorityQueueStateType(EmbeddedRocksDBStateBackend.PriorityQueueStateType priorityQueueStateType)
      Sets the type of the priority queue state. It will fallback to the default value, if it is not explicitly set.

    • setPredefinedOptions

      public void setPredefinedOptions(@Nonnull PredefinedOptions options)
      Sets the predefined options for RocksDB.

      If user-configured options within RocksDBConfigurableOptions is set (through config.yaml) or a user-defined options factory is set (via setRocksDBOptions(RocksDBOptionsFactory)), then the options from the factory are applied on top of the here specified predefined options and customized options.

      Parameters:
      options - The options to set (must not be null).

    • getPredefinedOptions

      @VisibleForTesting public PredefinedOptions getPredefinedOptions()
      Gets the currently set predefined options for RocksDB. The default options (if nothing was set via setPredefinedOptions(PredefinedOptions)) are PredefinedOptions.DEFAULT.

      If user-configured options within RocksDBConfigurableOptions is set (through config.yaml) of a user-defined options factory is set (via setRocksDBOptions(RocksDBOptionsFactory)), then the options from the factory are applied on top of the predefined and customized options.

      Returns:
      The currently set predefined options for RocksDB.

    • setRocksDBOptions

      public void setRocksDBOptions(RocksDBOptionsFactory optionsFactory)
      Sets Options for the RocksDB instances. Because the options are not serializable and hold native code references, they must be specified through a factory.

      The options created by the factory here are applied on top of the pre-defined options profile selected via setPredefinedOptions(PredefinedOptions) and user-configured options from configuration set by configure(ReadableConfig, ClassLoader) with keys in RocksDBConfigurableOptions.

      Parameters:
      optionsFactory - The options factory that lazily creates the RocksDB options.

    • getRocksDBOptions

      @Nullable public RocksDBOptionsFactory getRocksDBOptions()
      Gets Options for the RocksDB instances.

      The options created by the factory here are applied on top of the pre-defined options profile selected via setPredefinedOptions(PredefinedOptions). If the pre-defined options profile is the default (PredefinedOptions.DEFAULT), then the factory fully controls the RocksDB options.

    • getNumberOfTransferThreads

      public int getNumberOfTransferThreads()
      Gets the number of threads used to transfer files while snapshotting/restoring.

    • setNumberOfTransferThreads

      public void setNumberOfTransferThreads(int numberOfTransferThreads)
      Sets the number of threads used to transfer files while snapshotting/restoring.
      Parameters:
      numberOfTransferThreads - The number of threads used to transfer files while snapshotting/restoring.

    • getWriteBatchSize

      public long getWriteBatchSize()
      Gets the max batch size will be used in RocksDBWriteBatchWrapper.

    • setWriteBatchSize

      public void setWriteBatchSize(long writeBatchSize)
      Sets the max batch size will be used in RocksDBWriteBatchWrapper, no positive value will disable memory size controller, just use item count controller.
      Parameters:
      writeBatchSize - The size will used to be used in RocksDBWriteBatchWrapper.

    • setRocksDBMemoryFactory

      public void setRocksDBMemoryFactory(RocksDBMemoryControllerUtils.RocksDBMemoryFactory rocksDBMemoryFactory)
      Set RocksDBMemoryFactory.

    • toString

      public String toString()
      Overrides:
      toString in class Object