001/* 002 * Licensed to the Apache Software Foundation (ASF) under one 003 * or more contributor license agreements. See the NOTICE file 004 * distributed with this work for additional information 005 * regarding copyright ownership. The ASF licenses this file 006 * to you under the Apache License, Version 2.0 (the 007 * "License"); you may not use this file except in compliance 008 * with the License. You may obtain a copy of the License at 009 * 010 * http://www.apache.org/licenses/LICENSE-2.0 011 * 012 * Unless required by applicable law or agreed to in writing, software 013 * distributed under the License is distributed on an "AS IS" BASIS, 014 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 015 * See the License for the specific language governing permissions and 016 * limitations under the License. 017 */ 018package org.apache.hadoop.hbase.master; 019 020import static org.junit.Assert.assertEquals; 021import static org.junit.Assert.assertNotEquals; 022import static org.junit.Assert.assertTrue; 023 024import java.io.IOException; 025import org.apache.hadoop.conf.Configuration; 026import org.apache.hadoop.hbase.HBaseClassTestRule; 027import org.apache.hadoop.hbase.HBaseTestingUtility; 028import org.apache.hadoop.hbase.HRegionInfo; 029import org.apache.hadoop.hbase.MiniHBaseCluster; 030import org.apache.hadoop.hbase.MiniHBaseCluster.MiniHBaseClusterRegionServer; 031import org.apache.hadoop.hbase.ServerName; 032import org.apache.hadoop.hbase.Waiter; 033import org.apache.hadoop.hbase.master.assignment.RegionStates; 034import org.apache.hadoop.hbase.testclassification.MediumTests; 035import org.apache.hadoop.hbase.util.Bytes; 036import org.apache.hadoop.hbase.zookeeper.MetaTableLocator; 037import org.apache.hadoop.hbase.zookeeper.ZKUtil; 038import org.apache.hadoop.hbase.zookeeper.ZNodePaths; 039import org.apache.zookeeper.KeeperException; 040import org.junit.AfterClass; 041import org.junit.BeforeClass; 042import org.junit.ClassRule; 043import org.junit.Test; 044import org.junit.experimental.categories.Category; 045import org.slf4j.Logger; 046import org.slf4j.LoggerFactory; 047 048/** 049 * Tests handling of meta-carrying region server failover. 050 */ 051@Category(MediumTests.class) 052public class TestMetaShutdownHandler { 053 private static final Logger LOG = LoggerFactory.getLogger(TestMetaShutdownHandler.class); 054 @ClassRule 055 public static final HBaseClassTestRule CLASS_RULE = 056 HBaseClassTestRule.forClass(TestMetaShutdownHandler.class); 057 058 private final static HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility(); 059 final static Configuration conf = TEST_UTIL.getConfiguration(); 060 061 @BeforeClass 062 public static void setUpBeforeClass() throws Exception { 063 TEST_UTIL.startMiniCluster(1, 3, null, null, MyRegionServer.class); 064 } 065 066 @AfterClass 067 public static void tearDownAfterClass() throws Exception { 068 TEST_UTIL.shutdownMiniCluster(); 069 } 070 071 /** 072 * This test will test the expire handling of a meta-carrying 073 * region server. 074 * After HBaseMiniCluster is up, we will delete the ephemeral 075 * node of the meta-carrying region server, which will trigger 076 * the expire of this region server on the master. 077 * On the other hand, we will slow down the abort process on 078 * the region server so that it is still up during the master SSH. 079 * We will check that the master SSH is still successfully done. 080 */ 081 @Test 082 public void testExpireMetaRegionServer() throws Exception { 083 MiniHBaseCluster cluster = TEST_UTIL.getHBaseCluster(); 084 HMaster master = cluster.getMaster(); 085 RegionStates regionStates = master.getAssignmentManager().getRegionStates(); 086 ServerName metaServerName = regionStates.getRegionServerOfRegion( 087 HRegionInfo.FIRST_META_REGIONINFO); 088 if (master.getServerName().equals(metaServerName) || metaServerName == null 089 || !metaServerName.equals(cluster.getServerHoldingMeta())) { 090 // Move meta off master 091 metaServerName = 092 cluster.getLiveRegionServerThreads().get(0).getRegionServer().getServerName(); 093 master.move(HRegionInfo.FIRST_META_REGIONINFO.getEncodedNameAsBytes(), 094 Bytes.toBytes(metaServerName.getServerName())); 095 TEST_UTIL.waitUntilNoRegionsInTransition(60000); 096 metaServerName = regionStates.getRegionServerOfRegion(HRegionInfo.FIRST_META_REGIONINFO); 097 } 098 RegionState metaState = MetaTableLocator.getMetaRegionState(master.getZooKeeper()); 099 assertEquals("Wrong state for meta!", RegionState.State.OPEN, metaState.getState()); 100 assertNotEquals("Meta is on master!", metaServerName, master.getServerName()); 101 102 // Delete the ephemeral node of the meta-carrying region server. 103 // This is trigger the expire of this region server on the master. 104 String rsEphemeralNodePath = 105 ZNodePaths.joinZNode(master.getZooKeeper().getZNodePaths().rsZNode, 106 metaServerName.toString()); 107 ZKUtil.deleteNode(master.getZooKeeper(), rsEphemeralNodePath); 108 LOG.info("Deleted the znode for the RegionServer hosting hbase:meta; waiting on SSH"); 109 // Wait for SSH to finish 110 final ServerManager serverManager = master.getServerManager(); 111 final ServerName priorMetaServerName = metaServerName; 112 TEST_UTIL.waitFor(120000, 200, new Waiter.Predicate<Exception>() { 113 @Override 114 public boolean evaluate() throws Exception { 115 return !serverManager.isServerOnline(priorMetaServerName) 116 && !serverManager.areDeadServersInProgress(); 117 } 118 }); 119 LOG.info("Past wait on RIT"); 120 TEST_UTIL.waitUntilNoRegionsInTransition(60000); 121 // Now, make sure meta is assigned 122 assertTrue("Meta should be assigned", 123 regionStates.isRegionOnline(HRegionInfo.FIRST_META_REGIONINFO)); 124 // Now, make sure meta is registered in zk 125 metaState = MetaTableLocator.getMetaRegionState(master.getZooKeeper()); 126 assertEquals("Meta should not be in transition", RegionState.State.OPEN, 127 metaState.getState()); 128 assertEquals("Meta should be assigned", metaState.getServerName(), 129 regionStates.getRegionServerOfRegion(HRegionInfo.FIRST_META_REGIONINFO)); 130 assertNotEquals("Meta should be assigned on a different server", 131 metaState.getServerName(), metaServerName); 132 } 133 134 public static class MyRegionServer extends MiniHBaseClusterRegionServer { 135 136 public MyRegionServer(Configuration conf) throws IOException, KeeperException, 137 InterruptedException { 138 super(conf); 139 } 140 141 @Override 142 public void abort(String reason, Throwable cause) { 143 // sleep to slow down the region server abort 144 try { 145 Thread.sleep(30*1000); 146 } catch (InterruptedException e) { 147 return; 148 } 149 super.abort(reason, cause); 150 } 151 } 152}