Examples with DatasetConfigDTO com.linkedin.thirdeye.datalayer.dto.DatasetConfigDTO used on opensource projects

Example 1 with DatasetConfigDTO

use of com.linkedin.thirdeye.datalayer.dto.DatasetConfigDTO in project pinot by linkedin.

the class DetectionJobRunner method alignTimestampsToDataTimezone.

private DateTime alignTimestampsToDataTimezone(DateTime inputDateTime, String collection) {
    try {
        DatasetConfigDTO datasetConfig = DAO_REGISTRY.getDatasetConfigDAO().findByDataset(collection);
        TimeSpec timespec = ThirdEyeUtils.getTimeSpecFromDatasetConfig(datasetConfig);
        TimeGranularity dataGranularity = timespec.getDataGranularity();
        String timeFormat = timespec.getFormat();
        if (dataGranularity.getUnit().equals(TimeUnit.DAYS)) {
            DateTimeZone dataTimeZone = Utils.getDataTimeZone(collection);
            DateTimeFormatter inputDataDateTimeFormatter = DateTimeFormat.forPattern(timeFormat).withZone(dataTimeZone);
            long inputMillis = inputDateTime.getMillis();
            String inputDateTimeString = inputDataDateTimeFormatter.print(inputMillis);
            long timeZoneOffsetMillis = inputDataDateTimeFormatter.parseMillis(inputDateTimeString);
            inputDateTime = new DateTime(timeZoneOffsetMillis);
        }
    } catch (Exception e) {
        LOG.error("Exception in aligning timestamp to data time zone", e);
    }
    return inputDateTime;
}

Example 2 with DatasetConfigDTO

use of com.linkedin.thirdeye.datalayer.dto.DatasetConfigDTO in project pinot by linkedin.

the class DetectionJobScheduler method runAdhocAnomalyFunction.

/**
   * Point of entry for rest endpoints calling adhoc anomaly functions
   * TODO: Not updating detection status in case of adhoc currently, reconsider
   * @param functionId
   * @param startTime
   * @param endTime
   * @return job execution id
   */
public Long runAdhocAnomalyFunction(Long functionId, Long startTime, Long endTime) {
    Long jobExecutionId = null;
    AnomalyFunctionDTO anomalyFunction = DAO_REGISTRY.getAnomalyFunctionDAO().findById(functionId);
    String dataset = anomalyFunction.getCollection();
    DatasetConfigDTO datasetConfig = null;
    try {
        datasetConfig = CACHE_REGISTRY.getDatasetConfigCache().get(dataset);
    } catch (ExecutionException e) {
        LOG.error("Function: {} Dataset: {} Exception in fetching dataset config", functionId, dataset, e);
    }
    boolean pass = checkIfDetectionRunCriteriaMet(startTime, endTime, datasetConfig, anomalyFunction);
    if (pass) {
        jobExecutionId = runAnomalyFunctionOnRanges(anomalyFunction, Lists.newArrayList(startTime), Lists.newArrayList(endTime));
    } else {
        LOG.warn("Function: {} Dataset: {} Data incomplete for monitoring window {} ({}) to {} ({}), skipping anomaly detection", functionId, dataset, startTime, new DateTime(startTime), endTime, new DateTime(endTime));
    // TODO: Send email to owners/dev team
    }
    return jobExecutionId;
}

Example 3 with DatasetConfigDTO

use of com.linkedin.thirdeye.datalayer.dto.DatasetConfigDTO in project pinot by linkedin.

the class DatasetConfigCacheLoader method load.

@Override
public DatasetConfigDTO load(String collection) throws Exception {
    LOGGER.info("Loading DatasetConfigCache for {}", collection);
    DatasetConfigDTO datasetConfig = datasetConfigDAO.findByDataset(collection);
    return datasetConfig;
}

Example 4 with DatasetConfigDTO

use of com.linkedin.thirdeye.datalayer.dto.DatasetConfigDTO in project pinot by linkedin.

the class ThirdEyeCacheRegistry method initCaches.

private static void initCaches(ThirdEyeConfiguration config) {
    ThirdEyeCacheRegistry cacheRegistry = ThirdEyeCacheRegistry.getInstance();
    RemovalListener<PinotQuery, ResultSetGroup> listener = new RemovalListener<PinotQuery, ResultSetGroup>() {

        @Override
        public void onRemoval(RemovalNotification<PinotQuery, ResultSetGroup> notification) {
            LOGGER.info("Expired {}", notification.getKey().getPql());
        }
    };
    // ResultSetGroup Cache. The size of this cache is limited by the total number of buckets in all ResultSetGroup.
    // We estimate that 1 bucket (including overhead) consumes 1KB and this cache is allowed to use up to 50% of max
    // heap space.
    long maxBucketNumber = getApproximateMaxBucketNumber(DEFAULT_HEAP_PERCENTAGE_FOR_RESULTSETGROUP_CACHE);
    LoadingCache<PinotQuery, ResultSetGroup> resultSetGroupCache = CacheBuilder.newBuilder().removalListener(listener).expireAfterAccess(1, TimeUnit.HOURS).maximumWeight(maxBucketNumber).weigher((pinotQuery, resultSetGroup) -> {
        int resultSetCount = resultSetGroup.getResultSetCount();
        int weight = 0;
        for (int idx = 0; idx < resultSetCount; ++idx) {
            com.linkedin.pinot.client.ResultSet resultSet = resultSetGroup.getResultSet(idx);
            weight += (resultSet.getColumnCount() * resultSet.getRowCount());
        }
        return weight;
    }).build(new ResultSetGroupCacheLoader(pinotThirdeyeClientConfig));
    cacheRegistry.registerResultSetGroupCache(resultSetGroupCache);
    LOGGER.info("Max bucket number for ResultSetGroup cache is set to {}", maxBucketNumber);
    // CollectionMaxDataTime Cache
    LoadingCache<String, Long> collectionMaxDataTimeCache = CacheBuilder.newBuilder().refreshAfterWrite(5, TimeUnit.MINUTES).build(new CollectionMaxDataTimeCacheLoader(resultSetGroupCache, datasetConfigDAO));
    cacheRegistry.registerCollectionMaxDataTimeCache(collectionMaxDataTimeCache);
    // Query Cache
    QueryCache queryCache = new QueryCache(thirdEyeClient, Executors.newFixedThreadPool(10));
    cacheRegistry.registerQueryCache(queryCache);
    // Dimension Filter cache
    LoadingCache<String, String> dimensionFiltersCache = CacheBuilder.newBuilder().build(new DimensionFiltersCacheLoader(cacheRegistry.getQueryCache()));
    cacheRegistry.registerDimensionFiltersCache(dimensionFiltersCache);
    // Dashboards cache
    LoadingCache<String, String> dashboardsCache = CacheBuilder.newBuilder().build(new DashboardsCacheLoader(dashboardConfigDAO));
    cacheRegistry.registerDashboardsCache(dashboardsCache);
    // Collections cache
    CollectionsCache collectionsCache = new CollectionsCache(datasetConfigDAO, config);
    cacheRegistry.registerCollectionsCache(collectionsCache);
    // DatasetConfig cache
    LoadingCache<String, DatasetConfigDTO> datasetConfigCache = CacheBuilder.newBuilder().build(new DatasetConfigCacheLoader(datasetConfigDAO));
    cacheRegistry.registerDatasetConfigCache(datasetConfigCache);
    // MetricConfig cache
    LoadingCache<MetricDataset, MetricConfigDTO> metricConfigCache = CacheBuilder.newBuilder().build(new MetricConfigCacheLoader(metricConfigDAO));
    cacheRegistry.registerMetricConfigCache(metricConfigCache);
    // DashboardConfigs cache
    LoadingCache<String, List<DashboardConfigDTO>> dashboardConfigsCache = CacheBuilder.newBuilder().build(new DashboardConfigCacheLoader(dashboardConfigDAO));
    cacheRegistry.registerDashboardConfigsCache(dashboardConfigsCache);
}

Example 5 with DatasetConfigDTO

use of com.linkedin.thirdeye.datalayer.dto.DatasetConfigDTO in project pinot by linkedin.

the class AutoLoadPinotMetricsService method addNewDataset.

/**
   * Adds a new dataset to the thirdeye database
   * @param dataset
   * @param schema
   */
private void addNewDataset(String dataset, Schema schema) throws Exception {
    List<MetricFieldSpec> metricSpecs = schema.getMetricFieldSpecs();
    // Create DatasetConfig
    DatasetConfigDTO datasetConfigDTO = ConfigGenerator.generateDatasetConfig(dataset, schema);
    LOG.info("Creating dataset for {}", dataset);
    DAO_REGISTRY.getDatasetConfigDAO().save(datasetConfigDTO);
    // Create MetricConfig
    for (MetricFieldSpec metricFieldSpec : metricSpecs) {
        MetricConfigDTO metricConfigDTO = ConfigGenerator.generateMetricConfig(metricFieldSpec, dataset);
        LOG.info("Creating metric {} for {}", metricConfigDTO.getName(), dataset);
        DAO_REGISTRY.getMetricConfigDAO().save(metricConfigDTO);
    }
    // Create Default DashboardConfig
    List<Long> metricIds = ConfigGenerator.getMetricIdsFromMetricConfigs(DAO_REGISTRY.getMetricConfigDAO().findByDataset(dataset));
    DashboardConfigDTO dashboardConfigDTO = ConfigGenerator.generateDefaultDashboardConfig(dataset, metricIds);
    LOG.info("Creating default dashboard for dataset {}", dataset);
    DAO_REGISTRY.getDashboardConfigDAO().save(dashboardConfigDTO);
}