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.procedure; 019 020import static org.junit.Assert.assertEquals; 021import static org.junit.Assert.assertTrue; 022 023import java.io.IOException; 024import org.apache.hadoop.conf.Configuration; 025import org.apache.hadoop.hbase.HBaseTestingUtility; 026import org.apache.hadoop.hbase.MiniHBaseCluster; 027import org.apache.hadoop.hbase.ServerName; 028import org.apache.hadoop.hbase.TableName; 029import org.apache.hadoop.hbase.client.RegionInfo; 030import org.apache.hadoop.hbase.client.RegionReplicaTestHelper; 031import org.apache.hadoop.hbase.client.Table; 032import org.apache.hadoop.hbase.master.HMaster; 033import org.apache.hadoop.hbase.master.assignment.AssignmentTestingUtil; 034import org.apache.hadoop.hbase.procedure2.ProcedureExecutor; 035import org.apache.hadoop.hbase.procedure2.ProcedureTestingUtility; 036import org.junit.After; 037import org.junit.Before; 038import org.slf4j.Logger; 039import org.slf4j.LoggerFactory; 040 041public class TestSCPBase { 042 private static final Logger LOG = LoggerFactory.getLogger(TestSCPBase.class); 043 static final int RS_COUNT = 3; 044 045 protected HBaseTestingUtility util; 046 047 protected void setupConf(Configuration conf) { 048 conf.setInt(MasterProcedureConstants.MASTER_PROCEDURE_THREADS, 1); 049 } 050 051 @Before 052 public void setup() throws Exception { 053 this.util = new HBaseTestingUtility(); 054 setupConf(this.util.getConfiguration()); 055 startMiniCluster(); 056 ProcedureTestingUtility.setKillAndToggleBeforeStoreUpdate( 057 this.util.getHBaseCluster().getMaster().getMasterProcedureExecutor(), false); 058 } 059 060 protected void startMiniCluster() throws Exception { 061 this.util.startMiniCluster(RS_COUNT); 062 } 063 064 @After 065 public void tearDown() throws Exception { 066 MiniHBaseCluster cluster = this.util.getHBaseCluster(); 067 HMaster master = cluster == null ? null : cluster.getMaster(); 068 if (master != null && master.getMasterProcedureExecutor() != null) { 069 ProcedureTestingUtility.setKillAndToggleBeforeStoreUpdate(master.getMasterProcedureExecutor(), 070 false); 071 } 072 this.util.shutdownMiniCluster(); 073 } 074 075 /** 076 * Run server crash procedure steps twice to test idempotency and that we are persisting all 077 * needed state. 078 */ 079 protected void testRecoveryAndDoubleExecution(boolean carryingMeta, boolean doubleExecution) 080 throws Exception { 081 final TableName tableName = TableName.valueOf("testRecoveryAndDoubleExecution-carryingMeta-" 082 + carryingMeta + "-doubleExecution-" + doubleExecution); 083 try (Table t = createTable(tableName)) { 084 // Load the table with a bit of data so some logs to split and some edits in each region. 085 this.util.loadTable(t, HBaseTestingUtility.COLUMNS[0]); 086 final int count = util.countRows(t); 087 assertTrue("expected some rows", count > 0); 088 final String checksum = util.checksumRows(t); 089 // Run the procedure executor outside the master so we can mess with it. Need to disable 090 // Master's running of the server crash processing. 091 final HMaster master = this.util.getHBaseCluster().getMaster(); 092 final ProcedureExecutor<MasterProcedureEnv> procExec = master.getMasterProcedureExecutor(); 093 // find the first server that match the request and executes the test 094 ServerName rsToKill = null; 095 for (RegionInfo hri : util.getAdmin().getRegions(tableName)) { 096 final ServerName serverName = AssignmentTestingUtil.getServerHoldingRegion(util, hri); 097 if (AssignmentTestingUtil.isServerHoldingMeta(util, serverName) == carryingMeta) { 098 rsToKill = serverName; 099 break; 100 } 101 } 102 // Enable test flags and then queue the crash procedure. 103 ProcedureTestingUtility.waitNoProcedureRunning(procExec); 104 if (doubleExecution) { 105 // For SCP, if you enable this then we will enter an infinite loop, as we will crash between 106 // queue and open for TRSP, and then going back to queue, as we will use the crash rs as the 107 // target server since it is recored in hbase:meta. 108 ProcedureTestingUtility.setKillIfHasParent(procExec, false); 109 ProcedureTestingUtility.setKillAndToggleBeforeStoreUpdate(procExec, true); 110 // kill the RS 111 AssignmentTestingUtil.killRs(util, rsToKill); 112 long procId = getSCPProcId(procExec); 113 // Now run through the procedure twice crashing the executor on each step... 114 MasterProcedureTestingUtility.testRecoveryAndDoubleExecution(procExec, procId); 115 } else { 116 // kill the RS 117 AssignmentTestingUtil.killRs(util, rsToKill); 118 long procId = getSCPProcId(procExec); 119 ProcedureTestingUtility.waitProcedure(procExec, procId); 120 } 121 RegionReplicaTestHelper.assertReplicaDistributed(util, t); 122 assertEquals(count, util.countRows(t)); 123 assertEquals(checksum, util.checksumRows(t)); 124 } 125 } 126 127 protected long getSCPProcId(ProcedureExecutor<?> procExec) { 128 util.waitFor(30000, () -> !procExec.getProcedures().isEmpty()); 129 return procExec.getActiveProcIds().stream().mapToLong(Long::longValue).min().getAsLong(); 130 } 131 132 protected Table createTable(final TableName tableName) throws IOException { 133 final Table t = this.util.createTable(tableName, HBaseTestingUtility.COLUMNS, 134 HBaseTestingUtility.KEYS_FOR_HBA_CREATE_TABLE, getRegionReplication()); 135 return t; 136 } 137 138 protected int getRegionReplication() { 139 return 1; 140 } 141}