vanzin commented on a change in pull request #22138: [SPARK-25151][SS] Apply Apache Commons Pool to KafkaDataConsumer URL: https://github.com/apache/spark/pull/22138#discussion_r318275675
########## File path: external/kafka-0-10-sql/src/main/scala/org/apache/spark/sql/kafka010/FetchedDataPool.scala ########## @@ -0,0 +1,192 @@ +/* + * 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.kafka010 + +import java.{util => ju} +import java.util.concurrent.{ScheduledExecutorService, ScheduledFuture, TimeUnit} +import java.util.concurrent.atomic.LongAdder + +import scala.collection.mutable + +import org.apache.kafka.clients.consumer.ConsumerRecord + +import org.apache.spark.SparkEnv +import org.apache.spark.internal.Logging +import org.apache.spark.sql.kafka010.KafkaDataConsumer.{CacheKey, UNKNOWN_OFFSET} +import org.apache.spark.util.{Clock, SystemClock, ThreadUtils, Utils} + +/** + * Provides object pool for [[FetchedData]] which is grouped by [[CacheKey]]. + * + * Along with CacheKey, it receives desired start offset to find cached FetchedData which + * may be stored from previous batch. If it can't find one to match, it will create + * a new FetchedData. As "desired start offset" plays as second level of key which can be + * modified in same instance, this class cannot be replaced with general pool implementations + * including Apache Commons Pool which pools KafkaConsumer. + */ +private[kafka010] class FetchedDataPool( + executorService: ScheduledExecutorService, + clock: Clock) extends Logging { + import FetchedDataPool._ + + private val cache: mutable.Map[CacheKey, CachedFetchedDataList] = mutable.HashMap.empty + + private val (minEvictableIdleTimeMillis, evictorThreadRunIntervalMillis): (Long, Long) = { + val conf = SparkEnv.get.conf + + val minEvictIdleTime = conf.get(CONSUMER_CACHE_MIN_EVICTABLE_IDLE_TIME_MILLIS) + val evictorThreadInterval = conf.get(CONSUMER_CACHE_EVICTOR_THREAD_RUN_INTERVAL_MILLIS) + + (minEvictIdleTime, evictorThreadInterval) + } + + private def startEvictorThread(): ScheduledFuture[_] = { + executorService.scheduleAtFixedRate(new Runnable { + override def run(): Unit = { + Utils.tryLogNonFatalError(removeIdleFetchedData()) + } + }, 0, evictorThreadRunIntervalMillis, TimeUnit.MILLISECONDS) + } + + private var scheduled = startEvictorThread() + + private val numCreatedFetchedData = new LongAdder() + private val numTotalElements = new LongAdder() + + def numCreated: Long = numCreatedFetchedData.sum() + def numTotal: Long = numTotalElements.sum() + + def acquire(key: CacheKey, desiredStartOffset: Long): FetchedData = synchronized { + val fetchedDataList = cache.getOrElseUpdate(key, new CachedFetchedDataList()) + + val cachedFetchedDataOption = fetchedDataList.find { p => + !p.inUse && p.getObject.nextOffsetInFetchedData == desiredStartOffset + } + + var cachedFetchedData: CachedFetchedData = null + if (cachedFetchedDataOption.isDefined) { + cachedFetchedData = cachedFetchedDataOption.get + } else { + cachedFetchedData = CachedFetchedData.empty() + fetchedDataList += cachedFetchedData + + numCreatedFetchedData.increment() + numTotalElements.increment() + } + + cachedFetchedData.lastAcquiredTimestamp = clock.getTimeMillis() + cachedFetchedData.inUse = true + + cachedFetchedData.getObject + } + + def invalidate(key: CacheKey): Unit = synchronized { + cache.remove(key) match { + case Some(lst) => numTotalElements.add(-1 * lst.size) + case None => + } + } + + def release(key: CacheKey, fetchedData: FetchedData): Unit = synchronized { + def warnReleasedDataNotInPool(key: CacheKey, fetchedData: FetchedData): Unit = { + logWarning(s"No matching data in pool for $fetchedData in key $key. " + + "It might be released before, or it was not a part of pool.") + } + + cache.get(key) match { + case Some(fetchedDataList) => + val cachedFetchedDataOption = fetchedDataList.find { p => + p.inUse && p.getObject == fetchedData + } + + if (cachedFetchedDataOption.isEmpty) { + warnReleasedDataNotInPool(key, fetchedData) + } else { + val cachedFetchedData = cachedFetchedDataOption.get + cachedFetchedData.inUse = false + cachedFetchedData.lastReleasedTimestamp = clock.getTimeMillis() + } + + case None => + warnReleasedDataNotInPool(key, fetchedData) + } + } + + def shutdown(): Unit = { + ThreadUtils.shutdown(executorService) + } + + def reset(): Unit = synchronized { + scheduled.cancel(true) + + cache.clear() + numTotalElements.reset() + numCreatedFetchedData.reset() + + scheduled = startEvictorThread() + } + + private def removeIdleFetchedData(): Unit = synchronized { + val timestamp = clock.getTimeMillis() + val maxAllowedIdleTimestamp = timestamp - minEvictableIdleTimeMillis Review comment: maxAllowedReleasedTimestamp ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
