[
https://issues.apache.org/jira/browse/HIVE-26242?focusedWorklogId=785203&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-785203
]
ASF GitHub Bot logged work on HIVE-26242:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 27/Jun/22 18:28
Start Date: 27/Jun/22 18:28
Worklog Time Spent: 10m
Work Description: veghlaci05 commented on code in PR #3303:
URL: https://github.com/apache/hive/pull/3303#discussion_r907674601
##########
ql/src/java/org/apache/hadoop/hive/ql/txn/compactor/CompactionHeartbeatService.java:
##########
@@ -0,0 +1,222 @@
+/*
+ * 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.hadoop.hive.ql.txn.compactor;
+
+import org.apache.commons.pool2.ObjectPool;
+import org.apache.commons.pool2.impl.GenericObjectPool;
+import org.apache.commons.pool2.impl.GenericObjectPoolConfig;
+import org.apache.hadoop.hive.conf.HiveConf;
+import org.apache.hadoop.hive.metastore.IMetaStoreClient;
+import org.apache.hadoop.hive.metastore.conf.MetastoreConf;
+import org.apache.hive.common.util.ShutdownHookManager;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.util.HashMap;
+import java.util.NoSuchElementException;
+import java.util.Objects;
+import java.util.concurrent.Future;
+import java.util.concurrent.ScheduledThreadPoolExecutor;
+import java.util.concurrent.TimeUnit;
+
+import static
org.apache.hive.common.util.HiveStringUtils.SHUTDOWN_HOOK_PRIORITY;
+
+/**
+ * Singleton service responsible for heartbeating the compaction transactions.
+ */
+class CompactionHeartbeatService {
+
+
+ private static final Logger LOG =
LoggerFactory.getLogger(CompactionHeartbeatService.class);
+
+ private static volatile CompactionHeartbeatService instance;
+
+ /**
+ * Return the singleton instance of this class.
+ * @param conf The {@link HiveConf} used to create the service. Used only
during the firsst call
+ * @return Returns the singleton {@link CompactionHeartbeatService}
+ * @throws IllegalStateException Thrown when the service has already been
destroyed.
+ */
+ static CompactionHeartbeatService getInstance(HiveConf conf) {
+ if (instance == null) {
+ synchronized (CompactionHeartbeatService.class) {
+ if (instance == null) {
+ LOG.debug("Initializing compaction txn heartbeater service.");
+ instance = new CompactionHeartbeatService(conf);
+ ShutdownHookManager.addShutdownHook(() -> {
+ try {
+ instance.shutdown();
+ } catch (InterruptedException e) {
+ Thread.currentThread().interrupt();
+ }
+ }, SHUTDOWN_HOOK_PRIORITY);
+ }
+ }
+ }
+ if (instance.heartbeatExecutor.isShutdown()) {
+ throw new IllegalStateException("The CompactionHeartbeatService is
already destroyed!");
+ }
+ return instance;
+ }
+
+ private final ScheduledThreadPoolExecutor heartbeatExecutor;
+ private final ObjectPool<IMetaStoreClient> clientPool;
+ private final long initialDelay;
+ private final long period;
+ private final HashMap<Long, TaskWrapper> tasks = new HashMap<>(30);
+
+ /**
+ * Starts the heartbeat for the given transaction
+ * @param txnId The id of the compaction txn
+ * @param lockId The id of the lock associated with the txn
+ * @param tableName Required for logging only
+ * @throws IllegalStateException Thrown when the heartbeat for the given txn
has already been started.
+ */
+ void startHeartbeat(long txnId, long lockId, String tableName) {
+ if (tasks.containsKey(txnId)) {
+ throw new IllegalStateException("Heartbeat was already started for TXN "
+ txnId);
+ }
+ LOG.info("Submitting heartbeat task for TXN {}", txnId);
+ CompactionHeartbeater heartbeater = new CompactionHeartbeater(txnId,
lockId, tableName);
+ Future<?> submittedTask =
heartbeatExecutor.scheduleAtFixedRate(heartbeater, initialDelay, period,
TimeUnit.MILLISECONDS);
+ tasks.put(txnId, new TaskWrapper(heartbeater, submittedTask));
+ }
+
+ /**
+ * Stops the heartbeat for the given transaction
+ * @param txnId The id of the compaction txn
+ * @throws IllegalStateException Thrown when there is no {@link
CompactionHeartbeater} task associated with the
+ * given txnId.
+ */
+ void stopHeartbeat(long txnId) throws InterruptedException {
+ LOG.info("Stopping heartbeat task for TXN {}", txnId);
+ TaskWrapper wrapper = tasks.get(txnId);
+ if (wrapper == null) {
+ throw new IllegalStateException("No registered heartbeat found for TXN "
+ txnId);
+ }
+ wrapper.future.cancel(false);
+ try {
+ wrapper.heartbeater.waitUntilFinish(initialDelay);
+ } finally {
+ tasks.remove(txnId);
+ }
+ }
+
+ /**
+ * Shuts down the service, by closing its underlying resources. Be aware
that after shutdown this service is no
+ * longer usable, there is no way to re-initialize it.
+ * @throws InterruptedException
+ */
+ void shutdown() throws InterruptedException {
+ LOG.info("Shutting down compaction txn heartbeater service.");
+ heartbeatExecutor.shutdownNow();
+ try {
+ heartbeatExecutor.awaitTermination(5000, TimeUnit.MILLISECONDS);
+ } finally {
+ tasks.clear();
+ clientPool.close();
+ }
+ LOG.info("Compaction txn heartbeater service is successfully stopped.");
+ }
+
+ private CompactionHeartbeatService(HiveConf conf) {
+ int numberOfWorkers = MetastoreConf.getIntVar(conf,
MetastoreConf.ConfVars.COMPACTOR_WORKER_THREADS);
+ heartbeatExecutor = new ScheduledThreadPoolExecutor(0);
+ heartbeatExecutor.setRemoveOnCancelPolicy(true);
+ GenericObjectPoolConfig<IMetaStoreClient> config = new
GenericObjectPoolConfig<>();
+ config.setMinIdle(1);
+ config.setMaxIdle(2);
+ config.setMaxTotal(numberOfWorkers);
+ config.setBlockWhenExhausted(true);
+ config.setMaxWaitMillis(2000);
+ config.setTestOnBorrow(false);
+ config.setTestOnCreate(false);
+ config.setTestOnReturn(false);
+ config.setTestWhileIdle(false);
+ clientPool = new GenericObjectPool<>(new IMetaStoreClientFactory(conf),
config);
+ long txnTimeout = MetastoreConf.getTimeVar(conf,
MetastoreConf.ConfVars.TXN_TIMEOUT, TimeUnit.MILLISECONDS);
+ initialDelay = txnTimeout / 4;
+ period = txnTimeout / 2;
+ }
+
+ private final class CompactionHeartbeater implements Runnable {
+ final private Logger LOG =
LoggerFactory.getLogger(CompactionHeartbeater.class);
+ private final long txnId;
+ private final long lockId;
+ private final String tableName;
+ private final Object lock = new Object();
+ private volatile boolean running = false;
+
+
+ @Override
+ public void run() {
+ IMetaStoreClient msc = null;
+ try {
+ synchronized (lock) {
+ running = true;
+ }
+ LOG.debug("Heartbeating compaction transaction id {} for table: {}",
txnId, tableName);
+ // Create a metastore client for each thread since it is not thread
safe
+ msc = clientPool.borrowObject();
+ msc.heartbeat(txnId, lockId);
+ clientPool.returnObject(msc);
+ } catch (NoSuchElementException nsee) {
+ LOG.error("Compaction transaction heartbeater pool exhausted, unable
to heartbeat", nsee);
+ } catch (Exception e) {
+ LOG.error("Error while heartbeating compaction transaction id {} for
table: {}", txnId, tableName, e);
+ try {
+ clientPool.invalidateObject(msc);
+ } catch (Exception ex) {
+ LOG.error("Error while invalidating a broken MetaStoreClient
instance", e);
+ }
+ } finally {
+ synchronized (lock) {
+ running = false;
+ lock.notifyAll();
+ }
+ }
+ }
+
+ public void waitUntilFinish(long timeout) throws InterruptedException {
+ synchronized (lock) {
+ if (running) {
Review Comment:
If it's while it makes no sense to have timeout, we can possibly wait
forever. I think we should not block application shutdown at this point, if the
timeout is not enough for the heartbeater thread to finish, we should interrupt
it anyway.
Issue Time Tracking
-------------------
Worklog Id: (was: 785203)
Time Spent: 8h (was: 7h 50m)
> Compaction heartbeater improvements
> -----------------------------------
>
> Key: HIVE-26242
> URL: https://issues.apache.org/jira/browse/HIVE-26242
> Project: Hive
> Issue Type: Improvement
> Reporter: László Végh
> Assignee: László Végh
> Priority: Major
> Labels: pull-request-available
> Time Spent: 8h
> Remaining Estimate: 0h
>
> The Compaction heartbeater should be improved the following ways:
> * The metastore clients should be reused between heartbeats and closed only
> at the end, when the transaction ends
> * Instead of having a dedicated heartbeater thread for each Compaction
> transaction, there should be shared a heartbeater executor where the
> heartbeat tasks can be scheduled/submitted.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
