Examples with HadoopJarStep org.finra.herd.model.api.xml.HadoopJarStep used on opensource projects

Example 1 with HadoopJarStep

use of org.finra.herd.model.api.xml.HadoopJarStep in project herd by FINRAOS.

the class EmrDaoImpl method getStepConfig.

/**
 * Create the step config list of objects for hive/pig installation.
 *
 * @param emrClusterDefinition the EMR definition name value.
 *
 * @return list of step configuration that contains all the steps for the given configuration.
 */
private List<StepConfig> getStepConfig(EmrClusterDefinition emrClusterDefinition) {
    StepFactory stepFactory = new StepFactory();
    List<StepConfig> appSteps = new ArrayList<>();
    // Create install hive step and add to the StepConfig list
    if (StringUtils.isNotBlank(emrClusterDefinition.getHiveVersion())) {
        StepConfig installHive = new StepConfig().withName("Hive " + emrClusterDefinition.getHiveVersion()).withActionOnFailure(ActionOnFailure.TERMINATE_JOB_FLOW).withHadoopJarStep(stepFactory.newInstallHiveStep(emrClusterDefinition.getHiveVersion()));
        appSteps.add(installHive);
    }
    // Create install Pig step and add to the StepConfig List
    if (StringUtils.isNotBlank(emrClusterDefinition.getPigVersion())) {
        StepConfig installPig = new StepConfig().withName("Pig " + emrClusterDefinition.getPigVersion()).withActionOnFailure(ActionOnFailure.TERMINATE_JOB_FLOW).withHadoopJarStep(stepFactory.newInstallPigStep(emrClusterDefinition.getPigVersion()));
        appSteps.add(installPig);
    }
    // Add the hadoop jar steps that need to be added.
    if (!CollectionUtils.isEmpty(emrClusterDefinition.getHadoopJarSteps())) {
        for (HadoopJarStep hadoopJarStep : emrClusterDefinition.getHadoopJarSteps()) {
            StepConfig stepConfig = emrHelper.getEmrHadoopJarStepConfig(hadoopJarStep.getStepName(), hadoopJarStep.getJarLocation(), hadoopJarStep.getMainClass(), hadoopJarStep.getScriptArguments(), hadoopJarStep.isContinueOnError());
            appSteps.add(stepConfig);
        }
    }
    return appSteps;
}

Example 2 with HadoopJarStep

use of org.finra.herd.model.api.xml.HadoopJarStep in project herd by FINRAOS.

the class EmrServiceTest method testCreateEmrClusterStartupSteps.

/**
 * This method tests the cluster creation with the startup hadoop jar steps being added.
 */
@Test
public void testCreateEmrClusterStartupSteps() throws Exception {
    // Create the namespace entity.
    NamespaceEntity namespaceEntity = namespaceDaoTestHelper.createNamespaceEntity(NAMESPACE);
    String definitionXml = IOUtils.toString(resourceLoader.getResource(EMR_CLUSTER_DEFINITION_XML_FILE_WITH_CLASSPATH).getInputStream());
    EmrClusterDefinition expectedEmrClusterDefinition = xmlHelper.unmarshallXmlToObject(EmrClusterDefinition.class, definitionXml);
    emrClusterDefinitionDaoTestHelper.createEmrClusterDefinitionEntity(namespaceEntity, EMR_CLUSTER_DEFINITION_NAME, definitionXml);
    // Create a new EMR cluster create request
    EmrClusterCreateRequest request = getNewEmrClusterCreateRequest();
    EmrCluster emrCluster = emrService.createCluster(request);
    // Validate the returned object against the input.
    assertNotNull(emrCluster);
    assertTrue(emrCluster.getNamespace().equals(request.getNamespace()));
    assertTrue(emrCluster.getEmrClusterDefinitionName().equals(request.getEmrClusterDefinitionName()));
    assertTrue(emrCluster.getEmrClusterName().equals(request.getEmrClusterName()));
    assertNotNull(emrCluster.getId());
    assertNull(emrCluster.isDryRun());
    assertNotNull(emrCluster.getEmrClusterDefinition());
    assertTrue(emrCluster.isEmrClusterCreated());
    assertEquals(expectedEmrClusterDefinition, emrCluster.getEmrClusterDefinition());
    validateEmrClusterCreationLogUnique(emrCluster, expectedEmrClusterDefinition);
    ListStepsRequest listStepsRequest = (new ListStepsRequest()).withClusterId(emrCluster.getId());
    ListStepsResult listStepsResult = emrOperations.listStepsRequest(null, listStepsRequest);
    List<StepSummary> addedSteps = listStepsResult.getSteps();
    // Validate that the step was added.
    for (HadoopJarStep hadoopJarStep : expectedEmrClusterDefinition.getHadoopJarSteps()) {
        boolean stepFound = false;
        for (StepSummary stepSummary : addedSteps) {
            if (stepSummary.getName().equals(hadoopJarStep.getStepName())) {
                // Step found
                stepFound = true;
                break;
            }
        }
        assertTrue(stepFound);
    }
}

Example 3 with HadoopJarStep

use of org.finra.herd.model.api.xml.HadoopJarStep in project herd by FINRAOS.

the class EmrDaoTest method createEmrClusterAssertCallRunEmrJobFlowOptionalParams.

@Test
public void createEmrClusterAssertCallRunEmrJobFlowOptionalParams() throws Exception {
    String clusterName = "clusterName";
    EmrClusterDefinition emrClusterDefinition = new EmrClusterDefinition();
    InstanceDefinitions instanceDefinitions = new InstanceDefinitions();
    instanceDefinitions.setMasterInstances(new MasterInstanceDefinition(10, "masterInstanceType", NO_EMR_CLUSTER_DEFINITION_EBS_CONFIGURATION, NO_INSTANCE_SPOT_PRICE, NO_INSTANCE_MAX_SEARCH_PRICE, NO_INSTANCE_ON_DEMAND_THRESHOLD));
    instanceDefinitions.setCoreInstances(new InstanceDefinition(20, "coreInstanceType", NO_EMR_CLUSTER_DEFINITION_EBS_CONFIGURATION, BigDecimal.ONE, NO_INSTANCE_MAX_SEARCH_PRICE, NO_INSTANCE_ON_DEMAND_THRESHOLD));
    instanceDefinitions.setTaskInstances(new InstanceDefinition(30, "taskInstanceType", NO_EMR_CLUSTER_DEFINITION_EBS_CONFIGURATION, NO_INSTANCE_SPOT_PRICE, NO_INSTANCE_MAX_SEARCH_PRICE, NO_INSTANCE_ON_DEMAND_THRESHOLD));
    emrClusterDefinition.setInstanceDefinitions(instanceDefinitions);
    emrClusterDefinition.setNodeTags(Arrays.asList(new NodeTag("tagName", "tagValue"), new NodeTag("", "tagValue"), new NodeTag("tagName", "")));
    emrClusterDefinition.setSshKeyPairName("sshKeyPairName");
    emrClusterDefinition.setSubnetId("subnetId");
    emrClusterDefinition.setKeepAlive(true);
    emrClusterDefinition.setTerminationProtection(true);
    emrClusterDefinition.setHadoopVersion("hadoopVersion");
    emrClusterDefinition.setReleaseLabel("releaseLabel");
    emrClusterDefinition.setApplications(new ArrayList<>());
    {
        EmrClusterDefinitionApplication emrClusterDefinitionApplication = new EmrClusterDefinitionApplication();
        emrClusterDefinitionApplication.setName("applicationName1");
        emrClusterDefinitionApplication.setVersion("applicationVersion1");
        emrClusterDefinitionApplication.setArgs(Arrays.asList("applicationArg1"));
        emrClusterDefinition.getApplications().add(emrClusterDefinitionApplication);
    }
    {
        EmrClusterDefinitionApplication emrClusterDefinitionApplication = new EmrClusterDefinitionApplication();
        emrClusterDefinitionApplication.setName("applicationName2");
        emrClusterDefinitionApplication.setVersion("applicationVersion2");
        emrClusterDefinitionApplication.setArgs(Arrays.asList("applicationArg2"));
        emrClusterDefinitionApplication.setAdditionalInfoList(Arrays.asList(new Parameter("applicationAdditionalInfoName2", "applicationAdditionalInfoValue2")));
        emrClusterDefinition.getApplications().add(emrClusterDefinitionApplication);
    }
    emrClusterDefinition.setConfigurations(new ArrayList<>());
    {
        EmrClusterDefinitionConfiguration emrClusterDefinitionConfiguration = new EmrClusterDefinitionConfiguration();
        emrClusterDefinitionConfiguration.setClassification("classification");
        EmrClusterDefinitionConfiguration emrClusterDefinitionConfigurationInner = new EmrClusterDefinitionConfiguration();
        emrClusterDefinitionConfigurationInner.setClassification("classificationInner");
        emrClusterDefinitionConfiguration.setConfigurations(Arrays.asList(emrClusterDefinitionConfigurationInner));
        emrClusterDefinitionConfiguration.setProperties(Arrays.asList(new Parameter("propertyKey", "propertyValue")));
        emrClusterDefinition.getConfigurations().add(emrClusterDefinitionConfiguration);
    }
    emrClusterDefinition.setLogBucket("logBucket");
    emrClusterDefinition.setVisibleToAll(true);
    emrClusterDefinition.setEc2NodeIamProfileName("ec2NodeIamProfileName");
    emrClusterDefinition.setServiceIamRole("serviceIamRole");
    emrClusterDefinition.setAmiVersion("amiVersion");
    emrClusterDefinition.setAdditionalInfo("additionalInfo");
    emrClusterDefinition.setEncryptionEnabled(true);
    emrClusterDefinition.setDaemonConfigurations(Arrays.asList(new Parameter("daemonConfigurationsKey", "daemonConfigurationsValue")));
    ConfigurationFiles configurationFiles = new ConfigurationFiles();
    configurationFiles.getConfigurationFiles().add(new ConfigurationFile("fileNameShortcut", "configFileLocation"));
    KeyValuePairConfigurations keyValuePairConfigurations = new KeyValuePairConfigurations();
    keyValuePairConfigurations.getKeyValuePairConfigurations().add(new KeyValuePairConfiguration("keyValueShortcut", "attribKey", "attribVal"));
    emrClusterDefinition.setHadoopConfigurations(Arrays.asList(configurationFiles, keyValuePairConfigurations));
    emrClusterDefinition.setCustomBootstrapActionAll(new ArrayList<>());
    {
        ScriptDefinition scriptDefinitionAll = new ScriptDefinition();
        scriptDefinitionAll.setScriptName("scriptDefinitionAllName1");
        scriptDefinitionAll.setScriptLocation("scriptDefinitionAllLocation1");
        scriptDefinitionAll.setScriptArguments(Arrays.asList("scriptDefinitionAllArg1"));
        emrClusterDefinition.getCustomBootstrapActionAll().add(scriptDefinitionAll);
    }
    {
        ScriptDefinition scriptDefinitionAll = new ScriptDefinition();
        scriptDefinitionAll.setScriptName("scriptDefinitionAllName2");
        scriptDefinitionAll.setScriptLocation("scriptDefinitionAllLocation2");
        emrClusterDefinition.getCustomBootstrapActionAll().add(scriptDefinitionAll);
    }
    emrClusterDefinition.setCustomBootstrapActionMaster(new ArrayList<>());
    {
        ScriptDefinition scriptDefinitionMaster = new ScriptDefinition();
        scriptDefinitionMaster.setScriptName("scriptDefinitionMasterName1");
        scriptDefinitionMaster.setScriptLocation("scriptDefinitionMasterLocation1");
        scriptDefinitionMaster.setScriptArguments(Arrays.asList("scriptDefinitionMasterArg1"));
        emrClusterDefinition.getCustomBootstrapActionMaster().add(scriptDefinitionMaster);
    }
    {
        ScriptDefinition scriptDefinitionMaster = new ScriptDefinition();
        scriptDefinitionMaster.setScriptName("scriptDefinitionMasterName2");
        scriptDefinitionMaster.setScriptLocation("scriptDefinitionMasterLocation2");
        emrClusterDefinition.getCustomBootstrapActionMaster().add(scriptDefinitionMaster);
    }
    emrClusterDefinition.setHiveVersion("hiveVersion");
    emrClusterDefinition.setPigVersion("pigVersion");
    emrClusterDefinition.setInstallOozie(true);
    emrClusterDefinition.setHadoopJarSteps(Arrays.asList(new HadoopJarStep("stepName", "jarLocation", "mainClass", null, true)));
    emrClusterDefinition.setSupportedProduct("supportedProduct");
    emrClusterDefinition.setSecurityConfiguration("securityConfiguration");
    emrClusterDefinition.setScaleDownBehavior("scaleDownBehavior");
    emrClusterDefinition.setMasterSecurityGroup(EMR_MASTER_SECURITY_GROUP);
    emrClusterDefinition.setSlaveSecurityGroup(EMR_SLAVE_SECURITY_GROUP);
    String clusterId = "clusterId";
    when(mockEmrOperations.runEmrJobFlow(any(), any())).then(new Answer<String>() {

        @Override
        public String answer(InvocationOnMock invocation) throws Throwable {
            /*
                 * Assert that the given EMR cluster definition produced the correct RunJobFlowRequest
                 */
            RunJobFlowRequest runJobFlowRequest = invocation.getArgument(1);
            assertEquals(3, runJobFlowRequest.getInstances().getInstanceGroups().size());
            {
                InstanceGroupConfig instanceGroupConfig = runJobFlowRequest.getInstances().getInstanceGroups().get(1);
                assertEquals("1", instanceGroupConfig.getBidPrice());
            }
            {
                InstanceGroupConfig instanceGroupConfig = runJobFlowRequest.getInstances().getInstanceGroups().get(2);
                assertEquals("taskInstanceType", instanceGroupConfig.getInstanceType());
                assertEquals(30, instanceGroupConfig.getInstanceCount().intValue());
            }
            assertEquals("sshKeyPairName", runJobFlowRequest.getInstances().getEc2KeyName());
            assertEquals("subnetId", runJobFlowRequest.getInstances().getEc2SubnetId());
            assertEquals(true, runJobFlowRequest.getInstances().getKeepJobFlowAliveWhenNoSteps());
            assertEquals(true, runJobFlowRequest.getInstances().getTerminationProtected());
            assertEquals("hadoopVersion", runJobFlowRequest.getInstances().getHadoopVersion());
            assertEquals("releaseLabel", runJobFlowRequest.getReleaseLabel());
            assertEquals(2, runJobFlowRequest.getApplications().size());
            {
                Application application = runJobFlowRequest.getApplications().get(0);
                assertEquals("applicationName1", application.getName());
                assertEquals("applicationVersion1", application.getVersion());
                assertEquals(Arrays.asList("applicationArg1"), application.getArgs());
            }
            {
                Application application = runJobFlowRequest.getApplications().get(1);
                Map<String, String> additionalInfo = application.getAdditionalInfo();
                assertEquals(1, additionalInfo.size());
                assertEquals("applicationAdditionalInfoValue2", additionalInfo.get("applicationAdditionalInfoName2"));
            }
            assertEquals(1, runJobFlowRequest.getConfigurations().size());
            {
                Configuration configuration = runJobFlowRequest.getConfigurations().get(0);
                assertEquals("classification", configuration.getClassification());
                assertEquals(1, configuration.getConfigurations().size());
                {
                    Configuration configurationInner = configuration.getConfigurations().get(0);
                    assertEquals("classificationInner", configurationInner.getClassification());
                }
                assertEquals(1, configuration.getProperties().size());
                assertEquals("propertyValue", configuration.getProperties().get("propertyKey"));
            }
            assertEquals("logBucket", runJobFlowRequest.getLogUri());
            assertEquals(true, runJobFlowRequest.getVisibleToAllUsers());
            assertEquals("ec2NodeIamProfileName", runJobFlowRequest.getJobFlowRole());
            assertEquals("serviceIamRole", runJobFlowRequest.getServiceRole());
            assertEquals("amiVersion", runJobFlowRequest.getAmiVersion());
            assertEquals("additionalInfo", runJobFlowRequest.getAdditionalInfo());
            assertEquals(7, runJobFlowRequest.getBootstrapActions().size());
            {
                BootstrapActionConfig bootstrapActionConfig = runJobFlowRequest.getBootstrapActions().get(0);
                assertEquals("emr.encryption.script", bootstrapActionConfig.getName());
                ScriptBootstrapActionConfig scriptBootstrapAction = bootstrapActionConfig.getScriptBootstrapAction();
                assertEquals("s3:////herd_SCRIPTS/encrypt_disks.sh", scriptBootstrapAction.getPath());
                assertEquals(0, scriptBootstrapAction.getArgs().size());
            }
            {
                BootstrapActionConfig bootstrapActionConfig = runJobFlowRequest.getBootstrapActions().get(1);
                assertEquals("emr.aws.configure.daemon", bootstrapActionConfig.getName());
                ScriptBootstrapActionConfig scriptBootstrapAction = bootstrapActionConfig.getScriptBootstrapAction();
                assertEquals("s3://elasticmapreduce/bootstrap-actions/configure-daemons", scriptBootstrapAction.getPath());
                assertEquals(Arrays.asList("daemonConfigurationsKey=daemonConfigurationsValue"), scriptBootstrapAction.getArgs());
            }
            {
                BootstrapActionConfig bootstrapActionConfig = runJobFlowRequest.getBootstrapActions().get(2);
                assertEquals("emr.aws.configure.hadoop", bootstrapActionConfig.getName());
                ScriptBootstrapActionConfig scriptBootstrapAction = bootstrapActionConfig.getScriptBootstrapAction();
                assertEquals("s3://us-east-1.elasticmapreduce/bootstrap-actions/configure-hadoop", scriptBootstrapAction.getPath());
                assertEquals(Arrays.asList("fileNameShortcut", "configFileLocation", "keyValueShortcut", "attribKey=attribVal"), scriptBootstrapAction.getArgs());
            }
            {
                BootstrapActionConfig bootstrapActionConfig = runJobFlowRequest.getBootstrapActions().get(3);
                assertEquals("scriptDefinitionAllName1", bootstrapActionConfig.getName());
                ScriptBootstrapActionConfig scriptBootstrapAction = bootstrapActionConfig.getScriptBootstrapAction();
                assertEquals("scriptDefinitionAllLocation1", scriptBootstrapAction.getPath());
                assertEquals(Arrays.asList("scriptDefinitionAllArg1"), scriptBootstrapAction.getArgs());
            }
            {
                BootstrapActionConfig bootstrapActionConfig = runJobFlowRequest.getBootstrapActions().get(4);
                ScriptBootstrapActionConfig scriptBootstrapAction = bootstrapActionConfig.getScriptBootstrapAction();
                assertEquals(0, scriptBootstrapAction.getArgs().size());
            }
            {
                BootstrapActionConfig bootstrapActionConfig = runJobFlowRequest.getBootstrapActions().get(5);
                assertEquals("scriptDefinitionMasterName1", bootstrapActionConfig.getName());
                ScriptBootstrapActionConfig scriptBootstrapAction = bootstrapActionConfig.getScriptBootstrapAction();
                assertEquals("s3://elasticmapreduce/bootstrap-actions/run-if", scriptBootstrapAction.getPath());
                assertEquals(Arrays.asList("instance.isMaster=true", "scriptDefinitionMasterLocation1", "scriptDefinitionMasterArg1"), scriptBootstrapAction.getArgs());
            }
            {
                BootstrapActionConfig bootstrapActionConfig = runJobFlowRequest.getBootstrapActions().get(6);
                ScriptBootstrapActionConfig scriptBootstrapAction = bootstrapActionConfig.getScriptBootstrapAction();
                assertEquals(Arrays.asList("instance.isMaster=true", "scriptDefinitionMasterLocation2"), scriptBootstrapAction.getArgs());
            }
            assertEquals(Arrays.asList("supportedProduct"), runJobFlowRequest.getSupportedProducts());
            assertEquals("securityConfiguration", runJobFlowRequest.getSecurityConfiguration());
            assertEquals("scaleDownBehavior", runJobFlowRequest.getScaleDownBehavior());
            assertEquals(EMR_MASTER_SECURITY_GROUP, runJobFlowRequest.getInstances().getEmrManagedMasterSecurityGroup());
            assertEquals(EMR_SLAVE_SECURITY_GROUP, runJobFlowRequest.getInstances().getEmrManagedSlaveSecurityGroup());
            return clusterId;
        }
    });
    assertEquals(clusterId, emrDao.createEmrCluster(clusterName, emrClusterDefinition, new AwsParamsDto()));
}