viirya commented on a change in pull request #33763: URL: https://github.com/apache/spark/pull/33763#discussion_r697823961
########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/AvailableNowDataStreamWrapper.scala ########## @@ -0,0 +1,85 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.execution.streaming + +import org.apache.spark.internal.Logging +import org.apache.spark.sql.connector.read.streaming.{MicroBatchStream, ReadLimit, SparkDataStream, SupportsAdmissionControl, SupportsTriggerAvailableNow} +import org.apache.spark.sql.connector.read.streaming + +/** + * This class wraps a [[SparkDataStream]] and makes it support Trigger.AvailableNow, by + * overriding its [[latestOffset]] method to always return the latest offset when the method is + * first called. It is as if there is no new data coming in from the source after the first + * [[latestOffset]] call. + */ +class AvailableNowDataStreamWrapper(val delegate: SparkDataStream) + extends SparkDataStream with SupportsTriggerAvailableNow with Logging { + + private var fetchedOffset: Option[streaming.Offset] = _ + + override def initialOffset(): streaming.Offset = delegate.initialOffset() + + override def deserializeOffset(json: String): streaming.Offset = delegate.deserializeOffset(json) + + override def commit(end: streaming.Offset): Unit = delegate.commit(end) + + override def stop(): Unit = delegate.stop() + + override def prepareForTriggerAvailableNow(): Unit = {} Review comment: As this implements `SupportsTriggerAvailableNow`, but seems `prepareForTriggerAvailableNow` is useless? Why don't we record latest offset here? ########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/FileStreamSource.scala ########## @@ -105,6 +109,8 @@ class FileStreamSource( // Visible for testing and debugging in production. val seenFiles = new SeenFilesMap(maxFileAgeMs, fileNameOnly) + var allFilesForTriggerAvailableNow: Seq[(String, Long)] = null Review comment: `private`? ########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/MicroBatchExecution.scala ########## @@ -382,6 +404,22 @@ class MicroBatchExecution( // Generate a map from each unique source to the next available offset. val (nextOffsets, recentOffsets) = uniqueSources.toSeq.map { + case (s: AvailableNowDataStreamWrapper, limit) => + val originalSource = s.delegate + updateStatusMessage(s"Getting offsets from $s") + reportTimeTaken("latestOffset") { + val startOffsetOpt = availableOffsets.get(originalSource) + val startOffset = s match { Review comment: here shouldn't we match against `originalSource`? ########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/FileStreamSource.scala ########## @@ -126,7 +132,12 @@ class FileStreamSource( unreadFiles } else { // All the new files found - ignore aged files and files that we have seen. Review comment: Should this comment follow `fetchAllFiles`? ########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/MicroBatchExecution.scala ########## @@ -121,18 +122,36 @@ class MicroBatchExecution( // v2 source case r: StreamingDataSourceV2Relation => r.stream } - uniqueSources = sources.distinct.map { - case source: SupportsAdmissionControl => - val limit = source.getDefaultReadLimit - if (trigger == OneTimeTrigger && limit != ReadLimit.allAvailable()) { - logWarning(s"The read limit $limit for $source is ignored when Trigger.Once() is used.") - source -> ReadLimit.allAvailable() - } else { - source -> limit - } - case other => - other -> ReadLimit.allAvailable() - }.toMap + uniqueSources = triggerExecutor match { + case _: SingleBatchExecutor => + sources.distinct.map { + case s: SupportsAdmissionControl => + val limit = s.getDefaultReadLimit + if (limit != ReadLimit.allAvailable()) { + logWarning( + s"The read limit $limit for $s is ignored when Trigger.Once is used.") + } + s -> ReadLimit.allAvailable() + case s => + s -> ReadLimit.allAvailable() + }.toMap + + case _: MultiBatchExecutor => + sources.distinct.map { + case s: SupportsTriggerAvailableNow => s + case s: Source => new AvailableNowSourceWrapper(s) + case s: MicroBatchStream => new AvailableNowMicroBatchStreamWrapper(s) + }.map { s => + s.prepareForTriggerAvailableNow() + s -> s.getDefaultReadLimit Review comment: Should we log warning if it is `getDefaultReadLimit` is `ReadLimit.allAvailable()` for `AvailableNowTrigger`? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
