[
https://issues.apache.org/jira/browse/HIVE-21841?focusedWorklogId=261200&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-261200
]
ASF GitHub Bot logged work on HIVE-21841:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 17/Jun/19 07:04
Start Date: 17/Jun/19 07:04
Worklog Time Spent: 10m
Work Description: ashutosh-bapat commented on pull request #668:
HIVE-21841 : Add configuration to specify the hostname of HMS leader …to run
housekeeping tasks in.
URL: https://github.com/apache/hive/pull/668#discussion_r294158228
##########
File path:
itests/hive-unit/src/test/java/org/apache/hadoop/hive/metastore/MetastoreHousekeepingLeaderTestBase.java
##########
@@ -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.hadoop.hive.metastore;
+
+import org.apache.hadoop.conf.Configuration;
+import org.apache.hadoop.hdfs.MiniDFSCluster;
+import org.apache.hadoop.hive.metastore.conf.MetastoreConf;
+import org.apache.hadoop.hive.metastore.conf.MetastoreConf.ConfVars;
+import org.apache.hadoop.hive.metastore.security.HadoopThriftAuthBridge;
+import org.apache.hadoop.hive.ql.stats.StatsUpdaterThread;
+import org.apache.hadoop.hive.ql.txn.compactor.Cleaner;
+import org.apache.hadoop.hive.ql.txn.compactor.Initiator;
+import org.apache.hadoop.hive.ql.txn.compactor.Worker;
+import org.junit.Assert;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.util.HashMap;
+import java.util.Map;
+import java.util.Set;
+import java.util.concurrent.TimeUnit;
+
+/**
+ * Base class for HMS leader config testing.
+ */
+class MetastoreHousekeepingLeaderTestBase {
+ private static final Logger LOG =
LoggerFactory.getLogger(MetastoreHousekeepingLeaderTestBase.class);
+ private static HiveMetaStoreClient client;
+ private static Configuration conf = MetastoreConf.newMetastoreConf();
+ private static Warehouse warehouse;
+ private static boolean isServerStarted = false;
+ private static int port;
+ private static MiniDFSCluster miniDFS;
+ // How long should we wait for the housekeeping threads to start in ms.
+ private static final long SLEEP_INTERVAL_FOR_THREADS_TO_START = 10000;
+ // Threads using ThreadPool will start after the configured interval. So,
start them some time
+ // before we check the existence of threads.
+ private static final long REMOTE_TASKS_INTERVAL =
SLEEP_INTERVAL_FOR_THREADS_TO_START - 3000;
+ static final String METASTORE_THREAD_TASK_FREQ_CONF =
"metastore.leader.test.task.freq";
+
+ static Map<String, Boolean> threadNames = new HashMap<>();
+ static Map<Class, Boolean> threadClasses = new HashMap<>();
+
+ void internalSetup(final String leaderHostName) throws Exception {
+ MetaStoreTestUtils.setConfForStandloneMode(conf);
+ MetastoreConf.setVar(conf, ConfVars.THRIFT_BIND_HOST, "localhost");
+ MetastoreConf.setVar(conf,
ConfVars.METASTORE_HOUSEKEEPING_LEADER_HOSTNAME, leaderHostName);
+
+ addHouseKeepingThreadConfigs();
+
+ warehouse = new Warehouse(conf);
+
+ if (isServerStarted) {
+ Assert.assertNotNull("Unable to connect to the MetaStore server",
client);
+ return;
+ }
+
+ port =
MetaStoreTestUtils.startMetaStoreWithRetry(HadoopThriftAuthBridge.getBridge(),
+ conf, true);
+ System.out.println("Starting MetaStore Server on port " + port);
+ isServerStarted = true;
+
+ // If the client connects the metastore service has started. This is used
as a signal to
+ // start tests.
+ client = createClient();
+ }
+
+ private HiveMetaStoreClient createClient() throws Exception {
+ MetastoreConf.setVar(conf, ConfVars.THRIFT_URIS, "thrift://localhost:" +
port);
+ MetastoreConf.setBoolVar(conf, ConfVars.EXECUTE_SET_UGI, false);
+ return new HiveMetaStoreClient(conf);
+ }
+
+ private void addHouseKeepingThreadConfigs() throws Exception {
+ conf.setTimeDuration(METASTORE_THREAD_TASK_FREQ_CONF,
REMOTE_TASKS_INTERVAL,
+ TimeUnit.MILLISECONDS);
+ addStatsUpdaterThreadConfigs();
+ addReplChangeManagerConfigs();
+ addCompactorConfigs();
+ long numTasks = addRemoteOnlyTasksConfigs();
+ numTasks = numTasks + addAlwaysTasksConfigs();
+ MetastoreConf.setLongVar(conf, ConfVars.THREAD_POOL_SIZE, numTasks);
+ }
+ private void addStatsUpdaterThreadConfigs() {
+ MetastoreConf.setLongVar(conf, ConfVars.STATS_AUTO_UPDATE_WORKER_COUNT, 1);
+ MetastoreConf.setVar(conf, ConfVars.STATS_AUTO_UPDATE, "all");
+ threadClasses.put(StatsUpdaterThread.class, false);
+ threadNames.put(StatsUpdaterThread.WORKER_NAME_PREFIX, false);
+ }
+
+ private void addReplChangeManagerConfigs() throws Exception {
+ miniDFS = new MiniDFSCluster.Builder(new
Configuration()).numDataNodes(1).format(true).build();
+ MetastoreConf.setBoolVar(conf, ConfVars.REPLCMENABLED, true);
+ String cmroot = "hdfs://" + miniDFS.getNameNode().getHostAndPort() +
"/cmroot";
+ MetastoreConf.setVar(conf, ConfVars.REPLCMDIR, cmroot);
+ threadNames.put(ReplChangeManager.CM_THREAD_NAME_PREFIX, false);
+ }
+
+ private void addCompactorConfigs() {
+ MetastoreConf.setBoolVar(conf, ConfVars.COMPACTOR_INITIATOR_ON, true);
+ MetastoreConf.setVar(conf, ConfVars.HIVE_METASTORE_RUNWORKER_IN,
"metastore");
+ MetastoreConf.setLongVar(conf, ConfVars.COMPACTOR_WORKER_THREADS, 1);
+ threadClasses.put(Initiator.class, false);
+ threadClasses.put(Worker.class, false);
+ threadClasses.put(Cleaner.class, false);
+ }
+
+ private long addRemoteOnlyTasksConfigs() {
+ String remoteTaskClassPaths =
+ MetastoreTaskThreadTestImpl1.class.getCanonicalName() + "," +
+ MetastoreTaskThreadTestImpl2.class.getCanonicalName();
+
+ MetastoreConf.setVar(conf, ConfVars.TASK_THREADS_REMOTE_ONLY,
remoteTaskClassPaths);
+
+ threadNames.put(MetastoreTaskThreadTestImpl1.TASK_NAME, false);
+ threadNames.put(MetastoreTaskThreadTestImpl2.TASK_NAME, false);
+
+ return 2;
+ }
+
+ private long addAlwaysTasksConfigs() throws Exception {
+ String alwaysTaskClassPaths =
MetastoreTaskThreadTestImpl3.class.getCanonicalName();
+ MetastoreConf.setVar(conf, ConfVars.TASK_THREADS_ALWAYS,
alwaysTaskClassPaths);
+ threadNames.put(MetastoreTaskThreadTestImpl3.TASK_NAME, false);
+ return 1;
+ }
+
+ private static String getAllThreadsAsString() {
+ Map<Thread, StackTraceElement[]> threadStacks = Thread.getAllStackTraces();
+ StringBuilder sb = new StringBuilder();
+ for (Map.Entry<Thread, StackTraceElement[]> entry :
threadStacks.entrySet()) {
+ Thread t = entry.getKey();
+ sb.append(System.lineSeparator());
+ sb.append("Name: ").append(t.getName()).append(" State:
").append(t.getState())
+ .append(" Class name: ").append(t.getClass().getCanonicalName());
+ }
+ return sb.toString();
+ }
+
+ void searchHousekeepingThreads() throws Exception {
+ // Client has been created so the metastore has started serving. Sleep for
few seconds for
+ // the housekeeping threads to start.
+ Thread.sleep(SLEEP_INTERVAL_FOR_THREADS_TO_START);
Review comment:
That's why I am adding a rather long delay like 10s. Usually the threads are
started within a second after the metastore is started.
I don't want to add a synchronization just for testing to check whether
threads are started.
We could add retry mechanism similar to HiveMetaStoreClient() which creates
a connection to metastore. But we are already calling HiveMetaStoreClient() to
ensure that metastore has started to serve (the condition after which threads
are started). So, I don't see much point in retry as well. A sufficiently long
delay should be enough.
Do you have any specific solution in mind?
----------------------------------------------------------------
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]
Issue Time Tracking
-------------------
Worklog Id: (was: 261200)
Time Spent: 3h (was: 2h 50m)
> Leader election in HMS to run housekeeping tasks.
> -------------------------------------------------
>
> Key: HIVE-21841
> URL: https://issues.apache.org/jira/browse/HIVE-21841
> Project: Hive
> Issue Type: New Feature
> Affects Versions: 4.0.0
> Reporter: Ashutosh Bapat
> Assignee: Ashutosh Bapat
> Priority: Major
> Labels: pull-request-available
> Fix For: 4.0.0
>
> Attachments: HIVE-21841.01.patch, HIVE-21841.02.patch
>
> Time Spent: 3h
> Remaining Estimate: 0h
>
> HMS performs housekeeping tasks. When there are multiple HMSes we need to
> have a leader HMS elected which will carry out those housekeeping tasks.
> These tasks include execution of compaction tasks, auto-discovering
> partitions for external tables, generation of compaction tasks, repl thread
> etc.
> Note that, though the code for compaction tasks, auto-discovery of partitions
> etc. is in Hive, the actual tasks are initiated by an HMS configured to do
> so. So, leader election is required only for HMS and not for HS2.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
