Examples with JobType com.linkedin.drelephant.analysis.JobType used on opensource projects

Example 1 with JobType

use of com.linkedin.drelephant.analysis.JobType in project dr-elephant by linkedin.

the class Web method restSearchOptions.

/**
 * This returns the rest search options which are filled in the forms for the search page.
 * @return Returns the json object which should be filled in the search form.
 * return object:
 * <pre>
 *  *{
 *  "search-options": {
 *    "jobcategory": [
 *      {
 *        "name": "SPARK",
 *        "jobtypes": [
 *          {
 *            "name": "Spark"
 *          }
 *        ],
 *        "heuristics": [
 *          {
 *            "name": "Spark Configuration Best Practice"
 *          },
 *          {
 *            "name": "Spark Memory Limit"
 *          },
 *          {
 *            "name": "Spark Stage Runtime"
 *          },
 *          {
 *            "name": "Spark Job Runtime"
 *          },
 *          {
 *            "name": "Spark Executor Load Balance"
 *          },
 *          {
 *            "name": "Spark Event Log Limit"
 *          }
 *        ]
 *      },
 *      {
 *        "name": "MAPREDUCE",
 *        "jobtypes": [
 *          {
 *            "name": "Pig"
 *          },
 *          {
 *            "name": "Hive"
 *          },
 *          {
 *            "name": "Cascading"
 *          },
 *          {
 *            "name": "Voldemort"
 *          },
 *          {
 *            "name": "Kafka"
 *          },
 *          {
 *            "name": "HadoopJava"
 *          }
 *        ],
 *        "heuristics": [
 *          {
 *            "name": "Mapper Data Skew"
 *          },
 *          {
 *            "name": "Mapper GC"
 *          },
 *          {
 *            "name": "Mapper Time"
 *          },
 *          {
 *            "name": "Mapper Speed"
 *          },
 *          {
 *            "name": "Mapper Spill"
 *          },
 *          {
 *            "name": "Mapper Memory"
 *          },
 *          {
 *            "name": "Reducer Data Skew"
 *          },
 *          {
 *            "name": "Reducer GC"
 *          },
 *          {
 *            "name": "Reducer Time"
 *          },
 *          {
 *            "name": "Reducer Memory"
 *          },
 *          {
 *            "name": "Shuffle & Sort"
 *          },
 *          {
 *            "name": "Exception"
 *          }
 *        ]
 *      }
 *    ],
 *    "severities": [
 *      {
 *        "name": "Critical",
 *        "value": 4
 *      },
 *      {
 *        "name": "Severe",
 *        "value": 3
 *      },
 *      {
 *        "name": "Moderate",
 *        "value": 2
 *      },
 *      {
 *        "name": "Low",
 *        "value": 1
 *      },
 *      {
 *        "name": "None",
 *        "value": 0
 *      }
 *    ],
 *    "id": "search"
 *  }
 *}
 * </pre>
 */
public static Result restSearchOptions() {
    JsonObject searchOptions = new JsonObject();
    JsonArray jobCategory = new JsonArray();
    JsonArray severities = new JsonArray();
    Map<ApplicationType, List<JobType>> applicationTypeListMap = ElephantContext.instance().getAppTypeToJobTypes();
    for (ApplicationType key : applicationTypeListMap.keySet()) {
        JsonObject applicationType = new JsonObject();
        JsonArray jobTypes = new JsonArray();
        JsonArray heuristics = new JsonArray();
        for (JobType jobtype : applicationTypeListMap.get(key)) {
            JsonObject jobTypeNode = new JsonObject();
            jobTypeNode.addProperty(JsonKeys.NAME, jobtype.getName());
            jobTypes.add(jobTypeNode);
        }
        for (Heuristic heuristic : ElephantContext.instance().getHeuristicsForApplicationType(key)) {
            JsonObject heuristicNode = new JsonObject();
            heuristicNode.addProperty(JsonKeys.NAME, heuristic.getHeuristicConfData().getHeuristicName());
            heuristics.add(heuristicNode);
        }
        applicationType.addProperty(JsonKeys.NAME, key.getName());
        applicationType.add(JsonKeys.JOB_TYPES, jobTypes);
        applicationType.add(JsonKeys.HEURISTICS, heuristics);
        jobCategory.add(applicationType);
    }
    for (Severity severity : Severity.values()) {
        JsonObject severityObject = new JsonObject();
        severityObject.addProperty(JsonKeys.NAME, severity.getText());
        severityObject.addProperty(JsonKeys.VALUE, severity.getValue());
        severities.add(severityObject);
    }
    searchOptions.add(JsonKeys.JOB_CATEGORY, jobCategory);
    searchOptions.add(JsonKeys.SEVERITIES, severities);
    searchOptions.addProperty(JsonKeys.ID, "search");
    JsonObject parent = new JsonObject();
    parent.add(JsonKeys.SEARCH_OPTS, searchOptions);
    return ok(new Gson().toJson(parent));
}

Example 2 with JobType

use of com.linkedin.drelephant.analysis.JobType in project dr-elephant by linkedin.

the class ElephantContext method configureSupportedApplicationTypes.

/**
 * Decides what application types can be supported.
 *
 * An application type is supported if all the below are true.
 * 1. A Fetcher is defined in FetcherConf.xml for the application type.
 * 2. At least one Heuristic is configured in HeuristicConf.xml for the application type.
 * 3. At least one job type is configured in JobTypeConf.xml for the application type.
 */
private void configureSupportedApplicationTypes() {
    Set<ApplicationType> supportedTypes = Sets.intersection(_typeToFetcher.keySet(), _typeToHeuristics.keySet());
    supportedTypes = Sets.intersection(supportedTypes, _appTypeToJobTypes.keySet());
    supportedTypes = Sets.intersection(supportedTypes, _typeToAggregator.keySet());
    _typeToAggregator.keySet().retainAll(supportedTypes);
    _typeToFetcher.keySet().retainAll(supportedTypes);
    _typeToHeuristics.keySet().retainAll(supportedTypes);
    _appTypeToJobTypes.keySet().retainAll(supportedTypes);
    logger.info("Configuring ElephantContext...");
    for (ApplicationType type : supportedTypes) {
        _nameToType.put(type.getName(), type);
        List<String> classes = new ArrayList<String>();
        List<Heuristic> heuristics = _typeToHeuristics.get(type);
        for (Heuristic heuristic : heuristics) {
            classes.add(heuristic.getClass().getName());
        }
        List<JobType> jobTypes = _appTypeToJobTypes.get(type);
        logger.info("Supports " + type.getName() + " application type, using " + _typeToFetcher.get(type).toString() + " fetcher class with Heuristics [" + StringUtils.join(classes, ", ") + "] and following JobTypes [" + StringUtils.join(jobTypes, ", ") + "].");
    }
}

Example 3 with JobType

use of com.linkedin.drelephant.analysis.JobType in project dr-elephant by linkedin.

the class ElephantContext method matchJobType.

/**
 * Get the matched job type given a
 *
 * @param data The HadoopApplicationData to check
 * @return The matched job type
 */
public JobType matchJobType(HadoopApplicationData data) {
    if (data != null) {
        List<JobType> jobTypeList = _appTypeToJobTypes.get(data.getApplicationType());
        Properties jobProp = data.getConf();
        for (JobType type : jobTypeList) {
            if (type.matchType(jobProp)) {
                return type;
            }
        }
    }
    return null;
}

Example 4 with JobType

use of com.linkedin.drelephant.analysis.JobType in project dr-elephant by linkedin.

the class JobTypeConfiguration method parseJobTypeConfiguration.

private void parseJobTypeConfiguration(Element configuration) {
    Map<ApplicationType, JobType> defaultMap = new HashMap<ApplicationType, JobType>();
    NodeList nodes = configuration.getChildNodes();
    int n = 0;
    for (int i = 0; i < nodes.getLength(); i++) {
        Node node = nodes.item(i);
        if (node.getNodeType() == Node.ELEMENT_NODE) {
            n++;
            Element jobTypeNode = (Element) node;
            String jobTypeName;
            Node jobTypeNameNode = jobTypeNode.getElementsByTagName("name").item(0);
            if (jobTypeNameNode == null) {
                throw new RuntimeException("No tag 'jobtype' in jobtype " + n);
            }
            jobTypeName = jobTypeNameNode.getTextContent();
            if (jobTypeName.equals("")) {
                throw new RuntimeException("Empty tag 'jobtype' in jobtype " + n);
            }
            // Truncate jobtype length for db constraint
            if (jobTypeName.length() > TYPE_LEN_LIMIT) {
                logger.info("Truncate type " + jobTypeName.length());
                jobTypeName = jobTypeName.substring(0, TYPE_LEN_LIMIT);
            }
            String jobConfName;
            Node jobConfNameNode = jobTypeNode.getElementsByTagName("conf").item(0);
            if (jobConfNameNode == null) {
                throw new RuntimeException("No tag 'conf' in jobtype " + jobTypeName);
            }
            jobConfName = jobConfNameNode.getTextContent();
            if (jobConfName.equals("")) {
                throw new RuntimeException("Empty tag 'conf' in jobtype " + jobTypeName);
            }
            String jobConfValue;
            Node jobConfValueNode = jobTypeNode.getElementsByTagName("value").item(0);
            if (jobConfValueNode == null) {
                // Default regex. match any char one or more times
                jobConfValue = ".*";
            } else {
                jobConfValue = jobConfValueNode.getTextContent();
                if (jobConfValue.equals("")) {
                    jobConfValue = ".*";
                }
            }
            String appTypeName;
            Node appTypeNameNode = jobTypeNode.getElementsByTagName("applicationtype").item(0);
            if (appTypeNameNode == null) {
                throw new RuntimeException("No tag 'applicationtype' in jobtype " + jobTypeName);
            }
            appTypeName = appTypeNameNode.getTextContent();
            ApplicationType appType = new ApplicationType(appTypeName);
            boolean isDefault = jobTypeNode.getElementsByTagName("isDefault").item(0) != null;
            JobType newJobType = null;
            try {
                newJobType = new JobType(jobTypeName, jobConfName, jobConfValue);
            } catch (PatternSyntaxException e) {
                throw new RuntimeException("Error processing this pattern.  Pattern:" + jobConfValue + " jobtype:" + jobTypeName);
            }
            String newJobTypeStr = String.format("jobType:%s, for application type:%s, isDefault:%s, confName:%s, confValue:%s.", jobTypeName, appTypeName, isDefault, jobConfName, jobConfValue);
            logger.info("Loaded " + newJobTypeStr);
            if (isDefault) {
                if (defaultMap.containsKey(appType)) {
                    throw new RuntimeException("Each application type should have one and only one default job type. Duplicate default job type: " + newJobTypeStr + " for application type: " + appType.getName());
                } else {
                    defaultMap.put(appType, newJobType);
                }
            } else {
                List<JobType> jobTypes = getJobTypeList(appType);
                jobTypes.add(newJobType);
            }
        }
    }
    // Append default maps to the end of each job type list
    for (Map.Entry<ApplicationType, JobType> entry : defaultMap.entrySet()) {
        ApplicationType appType = entry.getKey();
        JobType jobType = entry.getValue();
        List<JobType> jobTypes = getJobTypeList(appType);
        jobTypes.add(jobType);
    }
    // Sanity check
    for (ApplicationType appType : _appTypeToJobTypeList.keySet()) {
        if (!defaultMap.containsKey(appType)) {
            throw new RuntimeException("Each application type should have one and only one default job type, there is" + " none for application type: " + appType.getName() + ". Use <isDefault/> to tag one.");
        }
    }
    Integer jobTypesSize = 0;
    for (List<JobType> jobTypes : _appTypeToJobTypeList.values()) {
        jobTypesSize += jobTypes.size();
    }
    logger.info("Loaded total " + jobTypesSize + " job types for " + _appTypeToJobTypeList.size() + " app types");
}