Github user HeartSaVioR commented on a diff in the pull request: https://github.com/apache/spark/pull/22138#discussion_r213866495 --- Diff: external/kafka-0-10-sql/src/main/scala/org/apache/spark/sql/kafka010/InternalKafkaConsumerPool.scala --- @@ -0,0 +1,260 @@ +/* + * 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.ConcurrentHashMap + +import org.apache.commons.pool2.{BaseKeyedPooledObjectFactory, PooledObject, SwallowedExceptionListener} +import org.apache.commons.pool2.impl.{DefaultEvictionPolicy, DefaultPooledObject, GenericKeyedObjectPool, GenericKeyedObjectPoolConfig} + +import org.apache.spark.SparkEnv +import org.apache.spark.internal.Logging +import org.apache.spark.sql.kafka010.InternalKafkaConsumerPool._ +import org.apache.spark.sql.kafka010.KafkaDataConsumer._ + +/** + * Provides object pool for [[InternalKafkaConsumer]] which is grouped by [[CacheKey]]. + * + * This class leverages [[GenericKeyedObjectPool]] internally, hence providing methods based on + * the class, and same contract applies: after using the borrowed object, you must either call + * returnObject() if the object is healthy to return to pool, or invalidateObject() if the object + * should be destroyed. + * + * The capacity of pool is determined by "spark.sql.kafkaConsumerCache.capacity" config value, and + * the pool will have reasonable default value if the value is not provided. + * + * This class guarantees that no caller will get pooled object once the object is borrowed and + * not yet returned, hence provide thread-safety usage of non-thread-safe [[InternalKafkaConsumer]] + * unless caller shares the object to multiple threads. + */ +private[kafka010] class InternalKafkaConsumerPool(objectFactory: ObjectFactory, + poolConfig: PoolConfig) { + + private lazy val pool = { + val internalPool = new GenericKeyedObjectPool[CacheKey, InternalKafkaConsumer]( + objectFactory, poolConfig) + internalPool.setSwallowedExceptionListener(CustomSwallowedExceptionListener) + internalPool + } + + /** + * Borrows [[InternalKafkaConsumer]] object from the pool. If there's no idle object for the key, + * the pool will create the [[InternalKafkaConsumer]] object. + * + * If the pool doesn't have idle object for the key and also exceeds the capacity, pool will try + * to clear some of idle objects. If it doesn't help getting empty space to create new object, + * it will throw [[NoSuchElementException]] immediately. + * + * Borrowed object must be returned by either calling returnObject or invalidateObject, otherwise + * the object will be kept in pool as active object. + */ + def borrowObject(key: CacheKey, kafkaParams: ju.Map[String, Object]): InternalKafkaConsumer = { + updateKafkaParamForKey(key, kafkaParams) + pool.borrowObject(key) + } + + /** Returns borrowed object to the pool. */ + def returnObject(intConsumer: InternalKafkaConsumer): Unit = { + pool.returnObject(extractCacheKey(intConsumer), intConsumer) + } + + /** Invalidates (destroy) borrowed object to the pool. */ + def invalidateObject(intConsumer: InternalKafkaConsumer): Unit = { + pool.invalidateObject(extractCacheKey(intConsumer), intConsumer) + } + + /** + * Invalidates current idle and active (borrowed) objects for the key. It ensure no invalidated + * object will be provided again via borrowObject. + * + * It doesn't mean the key will not be available: valid objects will be available via calling + * borrowObject afterwards. + */ + def invalidateKey(key: CacheKey): Unit = { + // invalidate all idle consumers for the key + pool.clear(key) + + pool.getNumActive() --- End diff -- My bad. Will remove.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org