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.command.wf; 019 020 import java.io.IOException; 021 import java.net.URI; 022 import java.net.URISyntaxException; 023 import java.util.Collection; 024 import java.util.HashMap; 025 import java.util.HashSet; 026 import java.util.List; 027 import java.util.Map; 028 import java.util.Set; 029 030 import org.apache.hadoop.conf.Configuration; 031 import org.apache.hadoop.fs.FileSystem; 032 import org.apache.hadoop.fs.Path; 033 import org.apache.oozie.ErrorCode; 034 import org.apache.oozie.WorkflowActionBean; 035 import org.apache.oozie.WorkflowJobBean; 036 import org.apache.oozie.client.OozieClient; 037 import org.apache.oozie.client.WorkflowAction; 038 import org.apache.oozie.client.WorkflowJob; 039 import org.apache.oozie.command.CommandException; 040 import org.apache.oozie.command.PreconditionException; 041 import org.apache.oozie.executor.jpa.JPAExecutorException; 042 import org.apache.oozie.executor.jpa.WorkflowActionDeleteJPAExecutor; 043 import org.apache.oozie.executor.jpa.WorkflowActionsGetForJobJPAExecutor; 044 import org.apache.oozie.executor.jpa.WorkflowJobGetJPAExecutor; 045 import org.apache.oozie.executor.jpa.WorkflowJobUpdateJPAExecutor; 046 import org.apache.oozie.service.DagXLogInfoService; 047 import org.apache.oozie.service.HadoopAccessorException; 048 import org.apache.oozie.service.HadoopAccessorService; 049 import org.apache.oozie.service.JPAService; 050 import org.apache.oozie.service.Services; 051 import org.apache.oozie.service.WorkflowAppService; 052 import org.apache.oozie.service.WorkflowStoreService; 053 import org.apache.oozie.util.ConfigUtils; 054 import org.apache.oozie.util.InstrumentUtils; 055 import org.apache.oozie.util.LogUtils; 056 import org.apache.oozie.util.ParamChecker; 057 import org.apache.oozie.util.PropertiesUtils; 058 import org.apache.oozie.util.XConfiguration; 059 import org.apache.oozie.util.XLog; 060 import org.apache.oozie.util.XmlUtils; 061 import org.apache.oozie.workflow.WorkflowApp; 062 import org.apache.oozie.workflow.WorkflowException; 063 import org.apache.oozie.workflow.WorkflowInstance; 064 import org.apache.oozie.workflow.WorkflowLib; 065 import org.apache.oozie.workflow.lite.NodeHandler; 066 067 /** 068 * This is a RerunXCommand which is used for rerunn. 069 * 070 */ 071 public class ReRunXCommand extends WorkflowXCommand<Void> { 072 private final String jobId; 073 private Configuration conf; 074 private final String authToken; 075 private final Set<String> nodesToSkip = new HashSet<String>(); 076 public static final String TO_SKIP = "TO_SKIP"; 077 private WorkflowJobBean wfBean; 078 private List<WorkflowActionBean> actions; 079 private JPAService jpaService; 080 081 private static final Set<String> DISALLOWED_DEFAULT_PROPERTIES = new HashSet<String>(); 082 private static final Set<String> DISALLOWED_USER_PROPERTIES = new HashSet<String>(); 083 084 static { 085 String[] badUserProps = { PropertiesUtils.DAYS, PropertiesUtils.HOURS, PropertiesUtils.MINUTES, 086 PropertiesUtils.KB, PropertiesUtils.MB, PropertiesUtils.GB, PropertiesUtils.TB, PropertiesUtils.PB, 087 PropertiesUtils.RECORDS, PropertiesUtils.MAP_IN, PropertiesUtils.MAP_OUT, PropertiesUtils.REDUCE_IN, 088 PropertiesUtils.REDUCE_OUT, PropertiesUtils.GROUPS }; 089 PropertiesUtils.createPropertySet(badUserProps, DISALLOWED_USER_PROPERTIES); 090 091 String[] badDefaultProps = { PropertiesUtils.HADOOP_USER}; 092 PropertiesUtils.createPropertySet(badUserProps, DISALLOWED_DEFAULT_PROPERTIES); 093 PropertiesUtils.createPropertySet(badDefaultProps, DISALLOWED_DEFAULT_PROPERTIES); 094 } 095 096 public ReRunXCommand(String jobId, Configuration conf, String authToken) { 097 super("rerun", "rerun", 1); 098 this.jobId = ParamChecker.notEmpty(jobId, "jobId"); 099 this.conf = ParamChecker.notNull(conf, "conf"); 100 this.authToken = ParamChecker.notEmpty(authToken, "authToken"); 101 } 102 103 /* (non-Javadoc) 104 * @see org.apache.oozie.command.XCommand#execute() 105 */ 106 @Override 107 protected Void execute() throws CommandException { 108 InstrumentUtils.incrJobCounter(getName(), 1, getInstrumentation()); 109 LogUtils.setLogInfo(wfBean, logInfo); 110 WorkflowInstance oldWfInstance = this.wfBean.getWorkflowInstance(); 111 WorkflowInstance newWfInstance; 112 113 WorkflowAppService wps = Services.get().get(WorkflowAppService.class); 114 try { 115 XLog.Info.get().setParameter(DagXLogInfoService.TOKEN, conf.get(OozieClient.LOG_TOKEN)); 116 WorkflowApp app = wps.parseDef(conf, authToken); 117 XConfiguration protoActionConf = wps.createProtoActionConf(conf, authToken, true); 118 WorkflowLib workflowLib = Services.get().get(WorkflowStoreService.class).getWorkflowLibWithNoDB(); 119 120 URI uri = new URI(conf.get(OozieClient.APP_PATH)); 121 HadoopAccessorService has = Services.get().get(HadoopAccessorService.class); 122 Configuration fsConf = has.createJobConf(uri.getAuthority()); 123 FileSystem fs = has.createFileSystem(wfBean.getUser(), uri, fsConf); 124 125 Path configDefault = null; 126 // app path could be a directory 127 Path path = new Path(uri.getPath()); 128 if (!fs.isFile(path)) { 129 configDefault = new Path(path, SubmitXCommand.CONFIG_DEFAULT); 130 } else { 131 configDefault = new Path(path.getParent(), SubmitXCommand.CONFIG_DEFAULT); 132 } 133 134 if (fs.exists(configDefault)) { 135 Configuration defaultConf = new XConfiguration(fs.open(configDefault)); 136 PropertiesUtils.checkDisallowedProperties(defaultConf, DISALLOWED_DEFAULT_PROPERTIES); 137 XConfiguration.injectDefaults(defaultConf, conf); 138 } 139 140 PropertiesUtils.checkDisallowedProperties(conf, DISALLOWED_USER_PROPERTIES); 141 142 // Resolving all variables in the job properties. This ensures the Hadoop Configuration semantics are preserved. 143 // The Configuration.get function within XConfiguration.resolve() works recursively to get the final value corresponding to a key in the map 144 // Resetting the conf to contain all the resolved values is necessary to ensure propagation of Oozie properties to Hadoop calls downstream 145 conf = ((XConfiguration) conf).resolve(); 146 147 try { 148 newWfInstance = workflowLib.createInstance(app, conf, jobId); 149 } 150 catch (WorkflowException e) { 151 throw new CommandException(e); 152 } 153 wfBean.setAppName(app.getName()); 154 wfBean.setProtoActionConf(protoActionConf.toXmlString()); 155 } 156 catch (WorkflowException ex) { 157 throw new CommandException(ex); 158 } 159 catch (IOException ex) { 160 throw new CommandException(ErrorCode.E0803, ex); 161 } 162 catch (HadoopAccessorException ex) { 163 throw new CommandException(ex); 164 } 165 catch (URISyntaxException ex) { 166 throw new CommandException(ErrorCode.E0711, ex.getMessage(), ex); 167 } 168 169 try { 170 for (int i = 0; i < actions.size(); i++) { 171 if (!nodesToSkip.contains(actions.get(i).getName())) { 172 jpaService.execute(new WorkflowActionDeleteJPAExecutor(actions.get(i).getId())); 173 LOG.info("Deleting Action[{0}] for re-run", actions.get(i).getId()); 174 } 175 else { 176 copyActionData(newWfInstance, oldWfInstance); 177 } 178 } 179 180 wfBean.setAppPath(conf.get(OozieClient.APP_PATH)); 181 wfBean.setConf(XmlUtils.prettyPrint(conf).toString()); 182 wfBean.setLogToken(conf.get(OozieClient.LOG_TOKEN, "")); 183 wfBean.setUser(conf.get(OozieClient.USER_NAME)); 184 String group = ConfigUtils.getWithDeprecatedCheck(conf, OozieClient.JOB_ACL, OozieClient.GROUP_NAME, null); 185 wfBean.setGroup(group); 186 wfBean.setExternalId(conf.get(OozieClient.EXTERNAL_ID)); 187 wfBean.setEndTime(null); 188 wfBean.setRun(wfBean.getRun() + 1); 189 wfBean.setStatus(WorkflowJob.Status.PREP); 190 wfBean.setWorkflowInstance(newWfInstance); 191 jpaService.execute(new WorkflowJobUpdateJPAExecutor(wfBean)); 192 } 193 catch (JPAExecutorException e) { 194 throw new CommandException(e); 195 } 196 197 return null; 198 } 199 200 /** 201 * Loading the Wfjob and workflow actions. Parses the config and adds the nodes that are to be skipped to the 202 * skipped node list 203 * 204 * @throws CommandException 205 */ 206 @Override 207 protected void eagerLoadState() throws CommandException { 208 super.eagerLoadState(); 209 try { 210 jpaService = Services.get().get(JPAService.class); 211 if (jpaService != null) { 212 this.wfBean = jpaService.execute(new WorkflowJobGetJPAExecutor(this.jobId)); 213 this.actions = jpaService.execute(new WorkflowActionsGetForJobJPAExecutor(this.jobId)); 214 } 215 else { 216 throw new CommandException(ErrorCode.E0610); 217 } 218 219 if (conf != null) { 220 if (conf.getBoolean(OozieClient.RERUN_FAIL_NODES, false) == false) { //Rerun with skipNodes 221 Collection<String> skipNodes = conf.getStringCollection(OozieClient.RERUN_SKIP_NODES); 222 for (String str : skipNodes) { 223 // trimming is required 224 nodesToSkip.add(str.trim()); 225 } 226 LOG.debug("Skipnode size :" + nodesToSkip.size()); 227 } 228 else { 229 for (WorkflowActionBean action : actions) { // Rerun from failed nodes 230 if (action.getStatus() == WorkflowAction.Status.OK) { 231 nodesToSkip.add(action.getName()); 232 } 233 } 234 LOG.debug("Skipnode size are to rerun from FAIL nodes :" + nodesToSkip.size()); 235 } 236 StringBuilder tmp = new StringBuilder(); 237 for (String node : nodesToSkip) { 238 tmp.append(node).append(","); 239 } 240 LOG.debug("SkipNode List :" + tmp); 241 } 242 } 243 catch (Exception ex) { 244 throw new CommandException(ErrorCode.E0603, ex); 245 } 246 } 247 248 /** 249 * Checks the pre-conditions that are required for workflow to recover - Last run of Workflow should be completed - 250 * The nodes that are to be skipped are to be completed successfully in the base run. 251 * 252 * @throws org.apache.oozie.command.CommandException,PreconditionException On failure of pre-conditions 253 */ 254 @Override 255 protected void eagerVerifyPrecondition() throws CommandException, PreconditionException { 256 super.eagerVerifyPrecondition(); 257 if (!(wfBean.getStatus().equals(WorkflowJob.Status.FAILED) 258 || wfBean.getStatus().equals(WorkflowJob.Status.KILLED) || wfBean.getStatus().equals( 259 WorkflowJob.Status.SUCCEEDED))) { 260 throw new CommandException(ErrorCode.E0805, wfBean.getStatus()); 261 } 262 Set<String> unmachedNodes = new HashSet<String>(nodesToSkip); 263 for (WorkflowActionBean action : actions) { 264 if (nodesToSkip.contains(action.getName())) { 265 if (!action.getStatus().equals(WorkflowAction.Status.OK) 266 && !action.getStatus().equals(WorkflowAction.Status.ERROR)) { 267 throw new CommandException(ErrorCode.E0806, action.getName()); 268 } 269 unmachedNodes.remove(action.getName()); 270 } 271 } 272 if (unmachedNodes.size() > 0) { 273 StringBuilder sb = new StringBuilder(); 274 String separator = ""; 275 for (String s : unmachedNodes) { 276 sb.append(separator).append(s); 277 separator = ","; 278 } 279 throw new CommandException(ErrorCode.E0807, sb); 280 } 281 } 282 283 /** 284 * Copys the variables for skipped nodes from the old wfInstance to new one. 285 * 286 * @param newWfInstance : Source WF instance object 287 * @param oldWfInstance : Update WF instance 288 */ 289 private void copyActionData(WorkflowInstance newWfInstance, WorkflowInstance oldWfInstance) { 290 Map<String, String> oldVars = new HashMap<String, String>(); 291 Map<String, String> newVars = new HashMap<String, String>(); 292 oldVars = oldWfInstance.getAllVars(); 293 for (String var : oldVars.keySet()) { 294 String actionName = var.split(WorkflowInstance.NODE_VAR_SEPARATOR)[0]; 295 if (nodesToSkip.contains(actionName)) { 296 newVars.put(var, oldVars.get(var)); 297 } 298 } 299 for (String node : nodesToSkip) { 300 // Setting the TO_SKIP variable to true. This will be used by 301 // SignalCommand and LiteNodeHandler to skip the action. 302 newVars.put(node + WorkflowInstance.NODE_VAR_SEPARATOR + TO_SKIP, "true"); 303 String visitedFlag = NodeHandler.getLoopFlag(node); 304 // Removing the visited flag so that the action won't be considered 305 // a loop. 306 if (newVars.containsKey(visitedFlag)) { 307 newVars.remove(visitedFlag); 308 } 309 } 310 newWfInstance.setAllVars(newVars); 311 } 312 313 /* (non-Javadoc) 314 * @see org.apache.oozie.command.XCommand#getEntityKey() 315 */ 316 @Override 317 public String getEntityKey() { 318 return this.jobId; 319 } 320 321 /* (non-Javadoc) 322 * @see org.apache.oozie.command.XCommand#isLockRequired() 323 */ 324 @Override 325 protected boolean isLockRequired() { 326 return true; 327 } 328 329 /* (non-Javadoc) 330 * @see org.apache.oozie.command.XCommand#loadState() 331 */ 332 @Override 333 protected void loadState() throws CommandException { 334 } 335 336 /* (non-Javadoc) 337 * @see org.apache.oozie.command.XCommand#verifyPrecondition() 338 */ 339 @Override 340 protected void verifyPrecondition() throws CommandException, PreconditionException { 341 } 342 }