Search in sources :

Example 1 with S3RifFile

use of gov.cms.bfd.pipeline.ccw.rif.extract.s3.S3RifFile in project beneficiary-fhir-data by CMSgov.

the class DataSetSubsetter method downloadDataSet.

/**
 * @param options the {@link ExtractionOptions} to use
 * @param dataSetS3KeyPrefix the S3 key prefix (i.e. directory) of the data set to download
 * @param downloadDirectory the Path to the directory to download the RIF files locally to
 * @return the {@link S3RifFile}s that comprise the full 1M beneficiary dummy data set
 */
private static List<RifFile> downloadDataSet(ExtractionOptions options, String dataSetS3KeyPrefix, Path downloadDirectory) {
    AmazonS3 s3Client = S3Utilities.createS3Client(options);
    TransferManager transferManager = TransferManagerBuilder.standard().withS3Client(s3Client).build();
    String dataSetPrefix = "data-random/" + dataSetS3KeyPrefix;
    String manifestSuffix = "1_manifest.xml";
    Path manifestDownloadPath = downloadDirectory.resolve(manifestSuffix);
    if (!Files.exists(manifestDownloadPath)) {
        String manifestKey = String.format("%s/%s", dataSetPrefix, manifestSuffix);
        Download manifestDownload = transferManager.download(options.getS3BucketName(), manifestKey, manifestDownloadPath.toFile());
        try {
            manifestDownload.waitForCompletion();
        } catch (AmazonClientException | InterruptedException e) {
            throw new RuntimeException(e);
        }
    }
    LOGGER.info("Manifest downloaded.");
    DataSetManifest dummyDataSetManifest;
    try {
        JAXBContext jaxbContext = JAXBContext.newInstance(DataSetManifest.class);
        Unmarshaller jaxbUnmarshaller = jaxbContext.createUnmarshaller();
        dummyDataSetManifest = (DataSetManifest) jaxbUnmarshaller.unmarshal(manifestDownloadPath.toFile());
    } catch (JAXBException e) {
        throw new UncheckedJaxbException(e);
    }
    List<RifFile> rifFiles = new ArrayList<>();
    for (DataSetManifestEntry manifestEntry : dummyDataSetManifest.getEntries()) {
        String dataSetFileKey = String.format("%s/%s", dataSetPrefix, manifestEntry.getName());
        Path dataSetFileDownloadPath = downloadDirectory.resolve(manifestEntry.getName());
        if (!Files.exists(dataSetFileDownloadPath)) {
            LOGGER.info("Downloading RIF file: '{}'...", manifestEntry.getName());
            Download dataSetFileDownload = transferManager.download(options.getS3BucketName(), dataSetFileKey, dataSetFileDownloadPath.toFile());
            try {
                dataSetFileDownload.waitForCompletion();
            } catch (AmazonClientException | InterruptedException e) {
                throw new RuntimeException(e);
            }
        }
        RifFile dataSetFile = new LocalRifFile(dataSetFileDownloadPath, manifestEntry.getType());
        rifFiles.add(dataSetFile);
    }
    transferManager.shutdownNow();
    LOGGER.info("Original RIF files ready.");
    return rifFiles;
}
Also used : Path(java.nio.file.Path) AmazonS3(com.amazonaws.services.s3.AmazonS3) TransferManager(com.amazonaws.services.s3.transfer.TransferManager) S3RifFile(gov.cms.bfd.pipeline.ccw.rif.extract.s3.S3RifFile) RifFile(gov.cms.bfd.model.rif.RifFile) LocalRifFile(gov.cms.bfd.pipeline.ccw.rif.extract.LocalRifFile) DataSetManifest(gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetManifest) LocalRifFile(gov.cms.bfd.pipeline.ccw.rif.extract.LocalRifFile) AmazonClientException(com.amazonaws.AmazonClientException) JAXBException(javax.xml.bind.JAXBException) ArrayList(java.util.ArrayList) JAXBContext(javax.xml.bind.JAXBContext) UncheckedJaxbException(gov.cms.bfd.sharedutils.exceptions.UncheckedJaxbException) Unmarshaller(javax.xml.bind.Unmarshaller) Download(com.amazonaws.services.s3.transfer.Download) DataSetManifestEntry(gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetManifest.DataSetManifestEntry)

Example 2 with S3RifFile

use of gov.cms.bfd.pipeline.ccw.rif.extract.s3.S3RifFile in project beneficiary-fhir-data by CMSgov.

the class CcwRifLoadJob method call.

/**
 * @see gov.cms.bfd.pipeline.sharedutils.PipelineJob#call()
 */
@Override
public PipelineJobOutcome call() throws Exception {
    LOGGER.debug("Scanning for data sets to process...");
    // Update the queue from S3.
    dataSetQueue.updatePendingDataSets();
    // If no manifest was found, we're done (until next time).
    if (dataSetQueue.isEmpty()) {
        LOGGER.debug(LOG_MESSAGE_NO_DATA_SETS);
        listener.noDataAvailable();
        return PipelineJobOutcome.NOTHING_TO_DO;
    }
    // We've found the oldest manifest.
    DataSetManifest manifestToProcess = dataSetQueue.getNextDataSetToProcess().get();
    LOGGER.info("Found data set to process: '{}'." + " There were '{}' total pending data sets and '{}' completed ones.", manifestToProcess.toString(), dataSetQueue.getPendingManifestsCount(), dataSetQueue.getCompletedManifestsCount().get());
    /*
     * We've got a data set to process. However, it might still be uploading
     * to S3, so we need to wait for that to complete before we start
     * processing it.
     */
    boolean alreadyLoggedWaitingEvent = false;
    while (!dataSetIsAvailable(manifestToProcess)) {
        /*
       * We're very patient here, so we keep looping, but it's prudent to
       * pause between each iteration. TODO should eventually time out,
       * once we know how long transfers might take
       */
        try {
            if (!alreadyLoggedWaitingEvent) {
                LOGGER.info("Data set not ready. Waiting for it to finish uploading...");
                alreadyLoggedWaitingEvent = true;
            }
            Thread.sleep(1000 * 1);
        } catch (InterruptedException e) {
            /*
         * Many Java applications use InterruptedExceptions to signal
         * that a thread should stop what it's doing ASAP. This app
         * doesn't, so this is unexpected, and accordingly, we don't
         * know what to do. Safest bet is to blow up.
         */
            throw new RuntimeException(e);
        }
    }
    /*
     * Huzzah! We've got a data set to process and we've verified it's all there
     * waiting for us in S3. Now convert it into a RifFilesEvent (containing a List
     * of asynchronously-downloading S3RifFiles.
     */
    LOGGER.info(LOG_MESSAGE_DATA_SET_READY);
    List<S3RifFile> rifFiles = manifestToProcess.getEntries().stream().map(manifestEntry -> new S3RifFile(appMetrics, manifestEntry, s3TaskManager.downloadAsync(manifestEntry))).collect(Collectors.toList());
    RifFilesEvent rifFilesEvent = new RifFilesEvent(manifestToProcess.getTimestamp(), new ArrayList<>(rifFiles));
    /*
     * To save time for the next data set, peek ahead at it. If it's available and
     * it looks like there's enough disk space, start downloading it early in the
     * background.
     */
    Optional<DataSetManifest> secondManifestToProcess = dataSetQueue.getSecondDataSetToProcess();
    if (secondManifestToProcess.isPresent() && dataSetIsAvailable(secondManifestToProcess.get())) {
        Path tmpdir = Paths.get(System.getProperty("java.io.tmpdir"));
        long usableFreeTempSpace;
        try {
            usableFreeTempSpace = Files.getFileStore(tmpdir).getUsableSpace();
        } catch (IOException e) {
            throw new UncheckedIOException(e);
        }
        if (usableFreeTempSpace >= (50 * GIGA)) {
            secondManifestToProcess.get().getEntries().stream().forEach(manifestEntry -> s3TaskManager.downloadAsync(manifestEntry));
        }
    }
    /*
     * Now we hand that off to the DataSetMonitorListener, to do the *real*
     * work of actually processing that data set. It's important that we
     * block until it's completed, in order to ensure that we don't end up
     * processing multiple data sets in parallel (which would lead to data
     * consistency problems).
     */
    listener.dataAvailable(rifFilesEvent);
    LOGGER.info(LOG_MESSAGE_DATA_SET_COMPLETE);
    /*
     * Now that the data set has been processed, we need to ensure that we
     * don't end up processing it again. We ensure this two ways: 1) we keep
     * a list of the data sets most recently processed, and 2) we rename the
     * S3 objects that comprise that data set. (#1 is required as S3
     * deletes/moves are only *eventually* consistent, so #2 may not take
     * effect right away.)
     */
    rifFiles.stream().forEach(f -> f.cleanupTempFile());
    dataSetQueue.markProcessed(manifestToProcess);
    s3TaskManager.submit(new DataSetMoveTask(s3TaskManager, options, manifestToProcess));
    return PipelineJobOutcome.WORK_DONE;
}
Also used : S3RifFile(gov.cms.bfd.pipeline.ccw.rif.extract.s3.S3RifFile) NullPipelineJobArguments(gov.cms.bfd.pipeline.sharedutils.NullPipelineJobArguments) LoggerFactory(org.slf4j.LoggerFactory) PipelineJob(gov.cms.bfd.pipeline.sharedutils.PipelineJob) ExtractionOptions(gov.cms.bfd.pipeline.ccw.rif.extract.ExtractionOptions) DataSetMonitorListener(gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetMonitorListener) ListObjectsV2Result(com.amazonaws.services.s3.model.ListObjectsV2Result) ArrayList(java.util.ArrayList) DataSetManifestEntry(gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetManifest.DataSetManifestEntry) S3TaskManager(gov.cms.bfd.pipeline.ccw.rif.extract.s3.task.S3TaskManager) HashSet(java.util.HashSet) ListObjectsV2Request(com.amazonaws.services.s3.model.ListObjectsV2Request) PipelineJobSchedule(gov.cms.bfd.pipeline.sharedutils.PipelineJobSchedule) DataSetMoveTask(gov.cms.bfd.pipeline.ccw.rif.extract.s3.task.DataSetMoveTask) Path(java.nio.file.Path) PipelineJobOutcome(gov.cms.bfd.pipeline.sharedutils.PipelineJobOutcome) MetricRegistry(com.codahale.metrics.MetricRegistry) Logger(org.slf4j.Logger) Files(java.nio.file.Files) Set(java.util.Set) IOException(java.io.IOException) Collectors(java.util.stream.Collectors) DataSetManifest(gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetManifest) RifFilesEvent(gov.cms.bfd.model.rif.RifFilesEvent) UncheckedIOException(java.io.UncheckedIOException) List(java.util.List) ChronoUnit(java.time.temporal.ChronoUnit) Paths(java.nio.file.Paths) DataSetManifestId(gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetManifest.DataSetManifestId) Optional(java.util.Optional) DataSetQueue(gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetQueue) Pattern(java.util.regex.Pattern) Path(java.nio.file.Path) DataSetManifest(gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetManifest) UncheckedIOException(java.io.UncheckedIOException) IOException(java.io.IOException) UncheckedIOException(java.io.UncheckedIOException) DataSetMoveTask(gov.cms.bfd.pipeline.ccw.rif.extract.s3.task.DataSetMoveTask) S3RifFile(gov.cms.bfd.pipeline.ccw.rif.extract.s3.S3RifFile) RifFilesEvent(gov.cms.bfd.model.rif.RifFilesEvent)

Example 3 with S3RifFile

use of gov.cms.bfd.pipeline.ccw.rif.extract.s3.S3RifFile in project beneficiary-fhir-data by CMSgov.

the class S3RifFile method cleanupTempFile.

/**
 * Removes the local temporary file that was used to cache this {@link S3RifFile}'s corresponding
 * S3 object data locally.
 */
public void cleanupTempFile() {
    LOGGER.debug("Cleaning up '{}'...", this);
    /*
     * We need to either cancel the download or wait for it to complete and then clean up the file.
     * However, canceling isn't a thread-safe operation (which is bonkers, but true), so we'll just
     * wait for completion.
     */
    try {
        ManifestEntryDownloadResult fileDownloadResult = waitForDownload();
        Files.deleteIfExists(fileDownloadResult.getLocalDownload());
    } catch (IOException e) {
        throw new UncheckedIOException(e);
    } catch (CancellationException e) {
        LOGGER.debug("Download was cancelled and can't be cleaned up.");
    }
    LOGGER.debug("Cleaned up '{}'.", this);
}
Also used : ManifestEntryDownloadResult(gov.cms.bfd.pipeline.ccw.rif.extract.s3.task.ManifestEntryDownloadTask.ManifestEntryDownloadResult) CancellationException(java.util.concurrent.CancellationException) UncheckedIOException(java.io.UncheckedIOException) IOException(java.io.IOException) UncheckedIOException(java.io.UncheckedIOException)

Aggregations

DataSetManifest (gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetManifest)2 DataSetManifestEntry (gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetManifest.DataSetManifestEntry)2 S3RifFile (gov.cms.bfd.pipeline.ccw.rif.extract.s3.S3RifFile)2 IOException (java.io.IOException)2 UncheckedIOException (java.io.UncheckedIOException)2 Path (java.nio.file.Path)2 ArrayList (java.util.ArrayList)2 AmazonClientException (com.amazonaws.AmazonClientException)1 AmazonS3 (com.amazonaws.services.s3.AmazonS3)1 ListObjectsV2Request (com.amazonaws.services.s3.model.ListObjectsV2Request)1 ListObjectsV2Result (com.amazonaws.services.s3.model.ListObjectsV2Result)1 Download (com.amazonaws.services.s3.transfer.Download)1 TransferManager (com.amazonaws.services.s3.transfer.TransferManager)1 MetricRegistry (com.codahale.metrics.MetricRegistry)1 RifFile (gov.cms.bfd.model.rif.RifFile)1 RifFilesEvent (gov.cms.bfd.model.rif.RifFilesEvent)1 ExtractionOptions (gov.cms.bfd.pipeline.ccw.rif.extract.ExtractionOptions)1 LocalRifFile (gov.cms.bfd.pipeline.ccw.rif.extract.LocalRifFile)1 DataSetManifestId (gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetManifest.DataSetManifestId)1 DataSetMonitorListener (gov.cms.bfd.pipeline.ccw.rif.extract.s3.DataSetMonitorListener)1