Package org.apache.beam.sdk.transforms.splittabledofn
@DefaultAnnotation(org.checkerframework.checker.nullness.qual.NonNull.class)
package org.apache.beam.sdk.transforms.splittabledofn
Defines utilities related to splittable
DoFn
.-
ClassDescriptionAn
OffsetRangeTracker
for tracking a growable offset range.Provides the estimated end offset of the range.HasDefaultTracker<RestrictionT extends HasDefaultTracker<RestrictionT,TrackerT>, TrackerT extends RestrictionTracker<RestrictionT, ?>> Interface for restrictions for which a default implementation ofDoFn.NewTracker
is available, depending only on the restriction itself.HasDefaultWatermarkEstimator<WatermarkEstimatorStateT,WatermarkEstimatorT extends WatermarkEstimator<WatermarkEstimatorStateT>> Interface for watermark estimator state for which a default implementation ofDoFn.NewWatermarkEstimator
is available, depending only on the watermark estimator state itself.ManualWatermarkEstimator<WatermarkEstimatorStateT>AWatermarkEstimator
which is controlled manually from within aDoFn
.ARestrictionTracker
for claiming offsets in anOffsetRange
in a monotonically increasing fashion.RestrictionTracker<RestrictionT,PositionT> Manages access to the restriction and keeps track of its claimed part for a splittableDoFn
.AllRestrictionTracker
s SHOULD implement this interface to improve auto-scaling and splitting performance.A representation for the amount of known completed and remaining work.RestrictionTracker.TruncateResult<RestrictionT>A representation of the truncate result.SplitResult<RestrictionT>A representation of a split result.TimestampObservingWatermarkEstimator<WatermarkEstimatorStateT>AWatermarkEstimator
that observes the timestamps of all records output from aDoFn
.WatermarkEstimator<WatermarkEstimatorStateT>AWatermarkEstimator
which is used for estimating output watermarks of a splittableDoFn
.A set ofWatermarkEstimator
s that users can use to advance the output watermark for their associatedsplittable DoFn
s.Concrete implementation of aManualWatermarkEstimator
.A watermark estimator that observes timestamps of records output from a DoFn reporting the timestamp of the last element seen as the current watermark.A watermark estimator that tracks wall time.