hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestAssignmentManagerMetrics.java

   1 /**
   2  * Licensed to the Apache Software Foundation (ASF) under one
   3  * or more contributor license agreements.  See the NOTICE file
   4  * distributed with this work for additional information
   5  * regarding copyright ownership.  The ASF licenses this file
   6  * to you under the Apache License, Version 2.0 (the
   7  * "License"); you may not use this file except in compliance
   8  * with the License.  You may obtain a copy of the License at
   9  *
  10  *     http://www.apache.org/licenses/LICENSE-2.0
  11  *
  12  * Unless required by applicable law or agreed to in writing, software
  13  * distributed under the License is distributed on an "AS IS" BASIS,
  14  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  15  * See the License for the specific language governing permissions and
  16  * limitations under the License.
  17  */
  18 package org.apache.hadoop.hbase.master;
  19
  20 import static org.junit.Assert.fail;
  21
  22 import java.io.IOException;
  23 import org.apache.hadoop.conf.Configuration;
  24 import org.apache.hadoop.hbase.CompatibilityFactory;
  25 import org.apache.hadoop.hbase.HBaseClassTestRule;
  26 import org.apache.hadoop.hbase.HBaseTestingUtility;
  27 import org.apache.hadoop.hbase.HConstants;
  28 import org.apache.hadoop.hbase.MiniHBaseCluster;
  29 import org.apache.hadoop.hbase.TableName;
  30 import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;
  31 import org.apache.hadoop.hbase.client.CoprocessorDescriptorBuilder;
  32 import org.apache.hadoop.hbase.client.Put;
  33 import org.apache.hadoop.hbase.client.Table;
  34 import org.apache.hadoop.hbase.client.TableDescriptor;
  35 import org.apache.hadoop.hbase.client.TableDescriptorBuilder;
  36 import org.apache.hadoop.hbase.coprocessor.CoprocessorHost;
  37 import org.apache.hadoop.hbase.master.assignment.AssignmentManager;
  38 import org.apache.hadoop.hbase.test.MetricsAssertHelper;
  39 import org.apache.hadoop.hbase.testclassification.MasterTests;
  40 import org.apache.hadoop.hbase.testclassification.MediumTests;
  41 import org.apache.hadoop.hbase.util.Bytes;
  42 import org.apache.hadoop.hbase.util.TableDescriptorChecker;
  43 import org.junit.AfterClass;
  44 import org.junit.BeforeClass;
  45 import org.junit.ClassRule;
  46 import org.junit.Rule;
  47 import org.junit.Test;
  48 import org.junit.experimental.categories.Category;
  49 import org.junit.rules.TestName;
  50 import org.slf4j.Logger;
  51 import org.slf4j.LoggerFactory;
  52
  53 @Category({ MasterTests.class, MediumTests.class })
  54 public class TestAssignmentManagerMetrics {
  55
  56   @ClassRule
  57   public static final HBaseClassTestRule CLASS_RULE =
  58       HBaseClassTestRule.forClass(TestAssignmentManagerMetrics.class);
  59
  60   private static final Logger LOG = LoggerFactory.getLogger(TestAssignmentManagerMetrics.class);
  61   private static final MetricsAssertHelper METRICS_HELPER = CompatibilityFactory
  62       .getInstance(MetricsAssertHelper.class);
  63
  64   private static MiniHBaseCluster CLUSTER;
  65   private static HMaster MASTER;
  66   private static HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
  67   private static final int MSG_INTERVAL = 1000;
  68
  69   @Rule
  70   public TestName name = new TestName();
  71
  72   @BeforeClass
  73   public static void startCluster() throws Exception {
  74     LOG.info("Starting cluster");
  75     Configuration conf = TEST_UTIL.getConfiguration();
  76
  77     // Disable sanity check for coprocessor
  78     conf.setBoolean(TableDescriptorChecker.TABLE_SANITY_CHECKS, false);
  79
  80     // set RIT stuck warning threshold to a small value
  81     conf.setInt(HConstants.METRICS_RIT_STUCK_WARNING_THRESHOLD, 20);
  82
  83     // set msgInterval to 1 second
  84     conf.setInt("hbase.regionserver.msginterval", MSG_INTERVAL);
  85
  86     // set tablesOnMaster to none
  87     conf.set("hbase.balancer.tablesOnMaster", "none");
  88
  89     // set client sync wait timeout to 5sec
  90     conf.setInt("hbase.client.sync.wait.timeout.msec", 5000);
  91     conf.setInt(HConstants.HBASE_CLIENT_RETRIES_NUMBER, 1);
  92     conf.setInt(HConstants.HBASE_CLIENT_OPERATION_TIMEOUT, 2500);
  93     // set a small interval for updating rit metrics
  94     conf.setInt(AssignmentManager.RIT_CHORE_INTERVAL_MSEC_CONF_KEY, MSG_INTERVAL);
  95     // set a small assign attempts for avoiding assert when retrying. (HBASE-20533)
  96     conf.setInt(AssignmentManager.ASSIGN_MAX_ATTEMPTS, 3);
  97
  98     // keep rs online so it can report the failed opens.
  99     conf.setBoolean(CoprocessorHost.ABORT_ON_ERROR_KEY, false);
 100     TEST_UTIL.startMiniCluster(1);
 101     CLUSTER = TEST_UTIL.getHBaseCluster();
 102     MASTER = CLUSTER.getMaster();
 103   }
 104
 105   @AfterClass
 106   public static void after() throws Exception {
 107     LOG.info("AFTER {} <= IS THIS NULL?", TEST_UTIL);
 108     TEST_UTIL.shutdownMiniCluster();
 109   }
 110
 111   @Test
 112   public void testRITAssignmentManagerMetrics() throws Exception {
 113     final TableName TABLENAME = TableName.valueOf(name.getMethodName());
 114     final byte[] FAMILY = Bytes.toBytes("family");
 115     try (Table table = TEST_UTIL.createTable(TABLENAME, FAMILY)){
 116       final byte[] row = Bytes.toBytes("row");
 117       final byte[] qualifier = Bytes.toBytes("qualifier");
 118       final byte[] value = Bytes.toBytes("value");
 119
 120       Put put = new Put(row);
 121       put.addColumn(FAMILY, qualifier, value);
 122       table.put(put);
 123
 124       // Sleep 3 seconds, wait for doMetrics chore catching up
 125       Thread.sleep(MSG_INTERVAL * 3);
 126
 127       // check the RIT is 0
 128       MetricsAssignmentManagerSource amSource =
 129           MASTER.getAssignmentManager().getAssignmentManagerMetrics().getMetricsProcSource();
 130
 131       METRICS_HELPER.assertGauge(MetricsAssignmentManagerSource.RIT_COUNT_NAME, 0, amSource);
 132       METRICS_HELPER.assertGauge(MetricsAssignmentManagerSource.RIT_COUNT_OVER_THRESHOLD_NAME, 0,
 133           amSource);
 134
 135       // alter table with a non-existing coprocessor
 136
 137       TableDescriptor htd = TableDescriptorBuilder.newBuilder(TABLENAME)
 138         .setColumnFamily(ColumnFamilyDescriptorBuilder.of(FAMILY))
 139         .setCoprocessor(CoprocessorDescriptorBuilder.newBuilder("com.foo.FooRegionObserver")
 140           .setJarPath("hdfs:///foo.jar")
 141           .setPriority(1001)
 142           .setProperty("arg1", "1")
 143           .setProperty("arg2", "2")
 144           .build())
 145         .build();
 146       try {
 147         TEST_UTIL.getAdmin().modifyTable(htd);
 148         fail("Expected region failed to open");
 149       } catch (IOException e) {
 150         // expected, the RS will crash and the assignment will spin forever waiting for a RS
 151         // to assign the region. the region will not go to FAILED_OPEN because in this case
 152         // we have just one RS and it will do one retry.
 153         LOG.info("Expected error", e);
 154       }
 155
 156       // Sleep 5 seconds, wait for doMetrics chore catching up
 157       // the rit count consists of rit and failed opens. see RegionInTransitionStat#update
 158       // Waiting for the completion of rit makes the assert stable.
 159       TEST_UTIL.waitUntilNoRegionsInTransition();
 160       Thread.sleep(MSG_INTERVAL * 5);
 161       METRICS_HELPER.assertGauge(MetricsAssignmentManagerSource.RIT_COUNT_NAME, 1, amSource);
 162       METRICS_HELPER.assertGauge(MetricsAssignmentManagerSource.RIT_COUNT_OVER_THRESHOLD_NAME, 1,
 163         amSource);
 164     }
 165   }
 166 }