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 */ 018 package org.apache.oozie.action.hadoop; 019 020 import java.io.IOException; 021 import java.io.StringReader; 022 import java.util.List; 023 024 import org.apache.hadoop.conf.Configuration; 025 import org.apache.hadoop.fs.Path; 026 import org.apache.hadoop.mapred.Counters; 027 import org.apache.hadoop.mapred.JobClient; 028 import org.apache.hadoop.mapred.JobConf; 029 import org.apache.hadoop.mapred.JobID; 030 import org.apache.hadoop.mapred.RunningJob; 031 import org.apache.oozie.action.ActionExecutorException; 032 import org.apache.oozie.client.WorkflowAction; 033 import org.apache.oozie.util.XConfiguration; 034 import org.apache.oozie.util.XLog; 035 import org.apache.oozie.util.XmlUtils; 036 import org.jdom.Element; 037 import org.jdom.Namespace; 038 import org.json.simple.JSONObject; 039 040 public class MapReduceActionExecutor extends JavaActionExecutor { 041 042 public static final String OOZIE_ACTION_EXTERNAL_STATS_WRITE = "oozie.action.external.stats.write"; 043 public static final String HADOOP_COUNTERS = "hadoop.counters"; 044 private XLog log = XLog.getLog(getClass()); 045 046 public MapReduceActionExecutor() { 047 super("map-reduce"); 048 } 049 050 @Override 051 protected List<Class> getLauncherClasses() { 052 List<Class> classes = super.getLauncherClasses(); 053 classes.add(LauncherMain.class); 054 classes.add(MapReduceMain.class); 055 classes.add(StreamingMain.class); 056 classes.add(PipesMain.class); 057 return classes; 058 } 059 060 @Override 061 protected String getLauncherMain(Configuration launcherConf, Element actionXml) { 062 String mainClass; 063 Namespace ns = actionXml.getNamespace(); 064 if (actionXml.getChild("streaming", ns) != null) { 065 mainClass = launcherConf.get(LauncherMapper.CONF_OOZIE_ACTION_MAIN_CLASS, StreamingMain.class.getName()); 066 } 067 else { 068 if (actionXml.getChild("pipes", ns) != null) { 069 mainClass = launcherConf.get(LauncherMapper.CONF_OOZIE_ACTION_MAIN_CLASS, PipesMain.class.getName()); 070 } 071 else { 072 mainClass = launcherConf.get(LauncherMapper.CONF_OOZIE_ACTION_MAIN_CLASS, MapReduceMain.class.getName()); 073 } 074 } 075 return mainClass; 076 } 077 078 @Override 079 Configuration setupLauncherConf(Configuration conf, Element actionXml, Path appPath, Context context) throws ActionExecutorException { 080 super.setupLauncherConf(conf, actionXml, appPath, context); 081 conf.setBoolean("mapreduce.job.complete.cancel.delegation.tokens", true); 082 return conf; 083 } 084 085 @Override 086 @SuppressWarnings("unchecked") 087 Configuration setupActionConf(Configuration actionConf, Context context, Element actionXml, Path appPath) 088 throws ActionExecutorException { 089 Namespace ns = actionXml.getNamespace(); 090 if (actionXml.getChild("streaming", ns) != null) { 091 Element streamingXml = actionXml.getChild("streaming", ns); 092 String mapper = streamingXml.getChildTextTrim("mapper", ns); 093 String reducer = streamingXml.getChildTextTrim("reducer", ns); 094 String recordReader = streamingXml.getChildTextTrim("record-reader", ns); 095 List<Element> list = (List<Element>) streamingXml.getChildren("record-reader-mapping", ns); 096 String[] recordReaderMapping = new String[list.size()]; 097 for (int i = 0; i < list.size(); i++) { 098 recordReaderMapping[i] = list.get(i).getTextTrim(); 099 } 100 list = (List<Element>) streamingXml.getChildren("env", ns); 101 String[] env = new String[list.size()]; 102 for (int i = 0; i < list.size(); i++) { 103 env[i] = list.get(i).getTextTrim(); 104 } 105 StreamingMain.setStreaming(actionConf, mapper, reducer, recordReader, recordReaderMapping, env); 106 } 107 else { 108 if (actionXml.getChild("pipes", ns) != null) { 109 Element pipesXml = actionXml.getChild("pipes", ns); 110 String map = pipesXml.getChildTextTrim("map", ns); 111 String reduce = pipesXml.getChildTextTrim("reduce", ns); 112 String inputFormat = pipesXml.getChildTextTrim("inputformat", ns); 113 String partitioner = pipesXml.getChildTextTrim("partitioner", ns); 114 String writer = pipesXml.getChildTextTrim("writer", ns); 115 String program = pipesXml.getChildTextTrim("program", ns); 116 PipesMain.setPipes(actionConf, map, reduce, inputFormat, partitioner, writer, program, appPath); 117 } 118 } 119 actionConf = super.setupActionConf(actionConf, context, actionXml, appPath); 120 return actionConf; 121 } 122 123 @Override 124 public void end(Context context, WorkflowAction action) throws ActionExecutorException { 125 super.end(context, action); 126 JobClient jobClient = null; 127 boolean exception = false; 128 try { 129 if (action.getStatus() == WorkflowAction.Status.OK) { 130 Element actionXml = XmlUtils.parseXml(action.getConf()); 131 JobConf jobConf = createBaseHadoopConf(context, actionXml); 132 jobClient = createJobClient(context, jobConf); 133 RunningJob runningJob = jobClient.getJob(JobID.forName(action.getExternalId())); 134 if (runningJob == null) { 135 throw new ActionExecutorException(ActionExecutorException.ErrorType.FAILED, "MR002", 136 "Unknown hadoop job [{0}] associated with action [{1}]. Failing this action!", action 137 .getExternalId(), action.getId()); 138 } 139 140 // TODO this has to be done in a better way 141 if (!runningJob.getJobName().startsWith("oozie:action:")) { 142 throw new ActionExecutorException(ActionExecutorException.ErrorType.FAILED, "MR001", 143 "ID swap should have happened in launcher job [{0}]", action.getExternalId()); 144 } 145 146 Counters counters = runningJob.getCounters(); 147 if (counters != null) { 148 ActionStats stats = new MRStats(counters); 149 String statsJsonString = stats.toJSON(); 150 context.setVar(HADOOP_COUNTERS, statsJsonString); 151 152 // If action stats write property is set to false by user or 153 // size of stats is greater than the maximum allowed size, 154 // do not store the action stats 155 if (Boolean.parseBoolean(evaluateConfigurationProperty(actionXml, 156 OOZIE_ACTION_EXTERNAL_STATS_WRITE, "false")) 157 && (statsJsonString.getBytes().length <= getMaxExternalStatsSize())) { 158 context.setExecutionStats(statsJsonString); 159 log.debug( 160 "Printing stats for Map-Reduce action as a JSON string : [{0}]" + statsJsonString); 161 } 162 } 163 else { 164 context.setVar(HADOOP_COUNTERS, ""); 165 XLog.getLog(getClass()).warn("Could not find Hadoop Counters for: [{0}]", action.getExternalId()); 166 } 167 } 168 } 169 catch (Exception ex) { 170 exception = true; 171 throw convertException(ex); 172 } 173 finally { 174 if (jobClient != null) { 175 try { 176 jobClient.close(); 177 } 178 catch (Exception e) { 179 if (exception) { 180 log.error("JobClient error: ", e); 181 } 182 else { 183 throw convertException(e); 184 } 185 } 186 } 187 } 188 } 189 190 // Return the value of the specified configuration property 191 private String evaluateConfigurationProperty(Element actionConf, String key, String defaultValue) throws ActionExecutorException { 192 try { 193 if (actionConf != null) { 194 Namespace ns = actionConf.getNamespace(); 195 Element e = actionConf.getChild("configuration", ns); 196 String strConf = XmlUtils.prettyPrint(e).toString(); 197 XConfiguration inlineConf = new XConfiguration(new StringReader(strConf)); 198 return inlineConf.get(key, defaultValue); 199 } 200 return ""; 201 } 202 catch (IOException ex) { 203 throw convertException(ex); 204 } 205 } 206 207 @SuppressWarnings("unchecked") 208 private JSONObject counterstoJson(Counters counters) { 209 210 if (counters == null) { 211 return null; 212 } 213 214 JSONObject groups = new JSONObject(); 215 for (String gName : counters.getGroupNames()) { 216 JSONObject group = new JSONObject(); 217 for (Counters.Counter counter : counters.getGroup(gName)) { 218 String cName = counter.getName(); 219 Long cValue = counter.getCounter(); 220 group.put(cName, cValue); 221 } 222 groups.put(gName, group); 223 } 224 return groups; 225 } 226 227 /** 228 * Return the sharelib name for the action. 229 * 230 * @return returns <code>streaming</code> if mapreduce-streaming action, <code>NULL</code> otherwise. 231 * @param actionXml 232 */ 233 @Override 234 protected String getDefaultShareLibName(Element actionXml) { 235 Namespace ns = actionXml.getNamespace(); 236 return (actionXml.getChild("streaming", ns) != null) ? "mapreduce-streaming" : null; 237 } 238 239 }