jerrypeng commented on code in PR #38517: URL: https://github.com/apache/spark/pull/38517#discussion_r1049017655
########## sql/core/src/test/scala/org/apache/spark/sql/execution/streaming/AsyncProgressTrackingMicroBatchExecutionSuite.scala: ########## @@ -0,0 +1,1865 @@ +/* + * 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 java.io.{File, OutputStream} +import java.util.concurrent.{CountDownLatch, Semaphore, ThreadPoolExecutor, TimeUnit} + +import scala.collection.mutable.ListBuffer + +import org.apache.hadoop.fs.Path +import org.scalatest.BeforeAndAfter +import org.scalatest.matchers.should.Matchers +import org.scalatest.time.{Seconds, Span} + +import org.apache.spark.sql._ +import org.apache.spark.sql.catalyst.streaming.WriteToStream +import org.apache.spark.sql.connector.read.streaming +import org.apache.spark.sql.execution.streaming.AsyncProgressTrackingMicroBatchExecution.{ASYNC_PROGRESS_TRACKING_CHECKPOINTING_INTERVAL_MS, ASYNC_PROGRESS_TRACKING_ENABLED, ASYNC_PROGRESS_TRACKING_OVERRIDE_SINK_SUPPORT_CHECK} +import org.apache.spark.sql.functions.{column, window} +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.streaming.{StreamingQuery, StreamingQueryException, StreamTest, Trigger} +import org.apache.spark.sql.streaming.util.StreamManualClock +import org.apache.spark.util.{Clock, Utils} + +class AsyncProgressTrackingMicroBatchExecutionSuite + extends StreamTest + with BeforeAndAfter + with Matchers { + + import testImplicits._ + + after { + sqlContext.streams.active.foreach(_.stop()) + } + + def getListOfFiles(dir: String): List[File] = { + val d = new File(dir) + if (d.exists && d.isDirectory) { + d.listFiles.filter(_.isFile).toList + } else { + List[File]() + } + } + + def waitPendingOffsetWrites(streamExecution: StreamExecution): Unit = { + assert(streamExecution.isInstanceOf[AsyncProgressTrackingMicroBatchExecution]) + eventually(timeout(Span(5, Seconds))) { + streamExecution + .asInstanceOf[AsyncProgressTrackingMicroBatchExecution] + .areWritesPendingOrInProgress() should be(false) + } + } + + def waitPendingPurges(streamExecution: StreamExecution): Unit = { + assert(streamExecution.isInstanceOf[AsyncProgressTrackingMicroBatchExecution]) + eventually(timeout(Span(5, Seconds))) { + streamExecution + .asInstanceOf[AsyncProgressTrackingMicroBatchExecution] + .arePendingAsyncPurge should be(false) + } + } + + // test the basic functionality i.e. happy path + test("async WAL commits happy path") { + val checkpointLocation = Utils.createTempDir(namePrefix = "streaming.metadata").getCanonicalPath + + val inputData = new MemoryStream[Int](id = 0, sqlContext = sqlContext) + val ds = inputData.toDF() + + val tableName = "test" + + def startQuery(): StreamingQuery = { + ds.writeStream + .format("memory") + .queryName(tableName) + .option(ASYNC_PROGRESS_TRACKING_ENABLED, true) + .option(ASYNC_PROGRESS_TRACKING_CHECKPOINTING_INTERVAL_MS, 0) + .option("checkpointLocation", checkpointLocation) + .start() + } + val query = startQuery() + val expected = new ListBuffer[Row]() + for (j <- 0 until 100) { + for (i <- 0 until 10) { + val v = i + (j * 10) + inputData.addData({ v }) + expected += Row(v) + } + query.processAllAvailable() + } + + checkAnswer( + spark.table(tableName), + expected.toSeq + ) + } + + test("async WAL commits recovery") { + val checkpointLocation = Utils.createTempDir(namePrefix = "streaming.metadata").getCanonicalPath + + val inputData = new MemoryStream[Int](id = 0, sqlContext = sqlContext) + val ds = inputData.toDF() + + var index = 0 + // to synchronize producing and consuming messages so that + // we can generate and read the desired number of batches + var countDownLatch = new CountDownLatch(10) + val sem = new Semaphore(1) + val data = new ListBuffer[Int]() + def startQuery(): StreamingQuery = { + ds.writeStream + .foreachBatch((ds: Dataset[Row], batchId: Long) => { + ds.collect.foreach((row: Row) => { + data += row.getInt(0) + }: Unit) + countDownLatch.countDown() + index += 1 + sem.release() + }) + .option(ASYNC_PROGRESS_TRACKING_ENABLED, true) + .option(ASYNC_PROGRESS_TRACKING_CHECKPOINTING_INTERVAL_MS, 0) + .option(ASYNC_PROGRESS_TRACKING_OVERRIDE_SINK_SUPPORT_CHECK, true) + .option("checkpointLocation", checkpointLocation) + .start() + } + var query = startQuery() + + for (i <- 0 until 10) { + sem.acquire() + inputData.addData({ i }) + } + + try { + countDownLatch.await(streamingTimeout.toMillis, TimeUnit.MILLISECONDS) + } finally { + query.stop() + } + + assert(index == 10) + data should equal(Array(0, 1, 2, 3, 4, 5, 6, 7, 8, 9)) + + countDownLatch = new CountDownLatch(10) + + /** + * Start the query again + */ + query = startQuery() + + for (i <- 10 until 20) { + sem.acquire() + inputData.addData({ i }) + } + + try { + countDownLatch.await(streamingTimeout.toMillis, TimeUnit.MILLISECONDS) + } finally { + query.stop() + } + + // convert data to set to deduplicate results + data.toSet should equal( + Array(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19).toSet + ) + } + + test("async WAL commits turn on and off") { + val inputData = new MemoryStream[Int](id = 0, sqlContext = sqlContext) + val ds = inputData.toDS() + + val checkpointLocation = Utils.createTempDir(namePrefix = "streaming.metadata").getCanonicalPath + + testStream( + ds, + extraOptions = Map( + ASYNC_PROGRESS_TRACKING_ENABLED -> "true", + ASYNC_PROGRESS_TRACKING_CHECKPOINTING_INTERVAL_MS -> "0" + ) + )( + AddData(inputData, 0), + StartStream(checkpointLocation = checkpointLocation), + CheckAnswer(0), + AddData(inputData, 1), + CheckAnswer(0, 1), + AddData(inputData, 2), + CheckAnswer(0, 1, 2), + Execute { q => + waitPendingOffsetWrites(q) + // make sure we have removed all pending commits + q.offsetLog.asInstanceOf[AsyncOffsetSeqLog].pendingAsyncOffsetWrite() should be(0) + }, + StopStream + ) + + // offsets should be logged + getListOfFiles(checkpointLocation + "/offsets") + .filter(file => !file.isHidden) + .map(file => file.getName.toInt) + .sorted should equal(Array(0, 1, 2)) + + getListOfFiles(checkpointLocation + "/commits") + .filter(file => !file.isHidden) + .map(file => file.getName.toInt) + .sorted should equal(Array(0, 1, 2)) + + /** + * Starting stream second time with async progress tracking turned off + */ + testStream(ds)( + // add new data + AddData(inputData, 3), + StartStream(checkpointLocation = checkpointLocation), + CheckNewAnswer(3), + AddData(inputData, 4), + CheckNewAnswer(4), + StopStream + ) + + // offsets should be logged + getListOfFiles(checkpointLocation + "/offsets") + .filter(file => !file.isHidden) + .map(file => file.getName.toInt) + .sorted should equal(Array(0, 1, 2, 3, 4)) + // commits for batch 2, 3, 4 should be logged + getListOfFiles(checkpointLocation + "/commits") + .filter(file => !file.isHidden) + .map(file => file.getName.toInt) + .sorted should equal(Array(0, 1, 2, 3, 4)) + + /** + * Starting stream third time with async progress tracking turned back on + */ + testStream( + ds, + extraOptions = Map( + ASYNC_PROGRESS_TRACKING_ENABLED -> "true", + ASYNC_PROGRESS_TRACKING_CHECKPOINTING_INTERVAL_MS -> "0" + ) + )( + // add new data + AddData(inputData, 5), + StartStream(checkpointLocation = checkpointLocation), + // no data needs to be replayed because commit log is on previously + CheckNewAnswer(5), + AddData(inputData, 6), + CheckNewAnswer(6), + Execute { q => + waitPendingOffsetWrites(q) + // make sure we have removed all pending commits + q.offsetLog.asInstanceOf[AsyncOffsetSeqLog].pendingAsyncOffsetWrite() should be(0) + }, + StopStream + ) + + // offsets should be logged + getListOfFiles(checkpointLocation + "/offsets") + .filter(file => !file.isHidden) + .map(file => file.getName.toInt) + .sorted should equal(Array(0, 1, 2, 3, 4, 5, 6)) + // no new commits should be logged since async offset commits are enabled + getListOfFiles(checkpointLocation + "/commits") + .filter(file => !file.isHidden) + .map(file => file.getName.toInt) + .sorted should equal(Array(0, 1, 2, 3, 4, 5, 6)) + + /** + * Starting stream fourth time with async progress tracking turned off + */ + testStream(ds)( + // add new data + AddData(inputData, 7), + StartStream(checkpointLocation = checkpointLocation), + CheckNewAnswer(7), + AddData(inputData, 8), + CheckNewAnswer(8), + StopStream + ) + + // offsets should be logged + getListOfFiles(checkpointLocation + "/offsets") + .filter(file => !file.isHidden) + .map(file => file.getName.toInt) + .sorted should equal(Array(0, 1, 2, 3, 4, 5, 6, 7, 8)) + // commits for batch 2, 3, 4, 6, 7, 8 should be logged + getListOfFiles(checkpointLocation + "/commits") + .filter(file => !file.isHidden) + .map(file => file.getName.toInt) + .sorted should equal(Array(0, 1, 2, 3, 4, 5, 6, 7, 8)) + } + + test("Fail with once trigger") { + val inputData = new MemoryStream[Int](id = 0, sqlContext = sqlContext) + val ds = inputData.toDF() + + val e = intercept[IllegalArgumentException] { + ds.writeStream + .format("noop") + .trigger(Trigger.Once()) + .option(ASYNC_PROGRESS_TRACKING_ENABLED, true) + .start() + } + e.getMessage should equal("Async progress tracking cannot be used with Once trigger") + } + + test("Fail with available now trigger") { + + val inputData = new MemoryStream[Int](id = 0, sqlContext = sqlContext) + val ds = inputData.toDF() + + val e = intercept[IllegalArgumentException] { + ds.writeStream + .format("noop") + .trigger(Trigger.AvailableNow()) + .option(ASYNC_PROGRESS_TRACKING_ENABLED, true) + .start() + } + e.getMessage should equal("Async progress tracking cannot be used with AvailableNow trigger") + } + + test("switching between async wal commit enabled and trigger once") { + val checkpointLocation = Utils.createTempDir(namePrefix = "streaming.metadata").getCanonicalPath + + val inputData = new MemoryStream[Int](id = 0, sqlContext = sqlContext) + val ds = inputData.toDF() + + var index = 0 + var countDownLatch = new CountDownLatch(10) + var sem = new Semaphore(1) + val data = new ListBuffer[Int]() + def startQuery( Review Comment: There are subtle differences for methods that have this name. I think its more readable if it is around the context of the test that uses it. -- 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]
