Examples with ResourceId org.apache.beam.sdk.io.fs.ResourceId used on opensource projects

Example 21 with ResourceId

use of org.apache.beam.sdk.io.fs.ResourceId in project beam by apache.

the class FakeJobService method startLoadJob.

@Override
public void startLoadJob(JobReference jobRef, JobConfigurationLoad loadConfig) throws InterruptedException, IOException {
    synchronized (allJobs) {
        Job job = new Job();
        job.setJobReference(jobRef);
        job.setConfiguration(new JobConfiguration().setLoad(loadConfig));
        job.setKind(" bigquery#job");
        job.setStatus(new JobStatus().setState("PENDING"));
        // the caller.
        if (loadConfig.getSourceUris().size() > 0) {
            ImmutableList.Builder<ResourceId> sourceFiles = ImmutableList.builder();
            ImmutableList.Builder<ResourceId> loadFiles = ImmutableList.builder();
            for (String filename : loadConfig.getSourceUris()) {
                sourceFiles.add(FileSystems.matchNewResource(filename, false));
                loadFiles.add(FileSystems.matchNewResource(filename + ThreadLocalRandom.current().nextInt(), false));
            }
            FileSystems.copy(sourceFiles.build(), loadFiles.build(), MoveOptions.StandardMoveOptions.IGNORE_MISSING_FILES);
            filesForLoadJobs.put(jobRef.getProjectId(), jobRef.getJobId(), loadFiles.build());
        }
        allJobs.put(jobRef.getProjectId(), jobRef.getJobId(), new JobInfo(job));
    }
}

Example 22 with ResourceId

use of org.apache.beam.sdk.io.fs.ResourceId in project beam by apache.

the class WriteOneFilePerWindow method expand.

@Override
public PDone expand(PCollection<String> input) {
    // filenamePrefix may contain a directory and a filename component. Pull out only the filename
    // component from that path for the PerWindowFiles.
    String prefix = "";
    ResourceId resource = FileBasedSink.convertToFileResourceIfPossible(filenamePrefix);
    if (!resource.isDirectory()) {
        prefix = verifyNotNull(resource.getFilename(), "A non-directory resource should have a non-null filename: %s", resource);
    }
    TextIO.Write write = TextIO.write().to(resource.getCurrentDirectory()).withFilenamePolicy(new PerWindowFiles(prefix)).withWindowedWrites();
    if (numShards != null) {
        write = write.withNumShards(numShards);
    }
    return input.apply(write);
}

Example 23 with ResourceId

use of org.apache.beam.sdk.io.fs.ResourceId in project beam by apache.

the class DataflowRunner method run.

@Override
public DataflowPipelineJob run(Pipeline pipeline) {
    logWarningIfPCollectionViewHasNonDeterministicKeyCoder(pipeline);
    if (containsUnboundedPCollection(pipeline)) {
        options.setStreaming(true);
    }
    replaceTransforms(pipeline);
    LOG.info("Executing pipeline on the Dataflow Service, which will have billing implications " + "related to Google Compute Engine usage and other Google Cloud Services.");
    List<DataflowPackage> packages = options.getStager().stageFiles();
    // Set a unique client_request_id in the CreateJob request.
    // This is used to ensure idempotence of job creation across retried
    // attempts to create a job. Specifically, if the service returns a job with
    // a different client_request_id, it means the returned one is a different
    // job previously created with the same job name, and that the job creation
    // has been effectively rejected. The SDK should return
    // Error::Already_Exists to user in that case.
    int randomNum = new Random().nextInt(9000) + 1000;
    String requestId = DateTimeFormat.forPattern("YYYYMMddHHmmssmmm").withZone(DateTimeZone.UTC).print(DateTimeUtils.currentTimeMillis()) + "_" + randomNum;
    // Try to create a debuggee ID. This must happen before the job is translated since it may
    // update the options.
    DataflowPipelineOptions dataflowOptions = options.as(DataflowPipelineOptions.class);
    maybeRegisterDebuggee(dataflowOptions, requestId);
    JobSpecification jobSpecification = translator.translate(pipeline, this, packages);
    Job newJob = jobSpecification.getJob();
    newJob.setClientRequestId(requestId);
    ReleaseInfo releaseInfo = ReleaseInfo.getReleaseInfo();
    String version = releaseInfo.getVersion();
    checkState(!version.equals("${pom.version}"), "Unable to submit a job to the Dataflow service with unset version ${pom.version}");
    System.out.println("Dataflow SDK version: " + version);
    newJob.getEnvironment().setUserAgent((Map) releaseInfo.getProperties());
    // must be verified.
    if (!isNullOrEmpty(options.getGcpTempLocation())) {
        newJob.getEnvironment().setTempStoragePrefix(dataflowOptions.getPathValidator().verifyPath(options.getGcpTempLocation()));
    }
    newJob.getEnvironment().setDataset(options.getTempDatasetId());
    newJob.getEnvironment().setExperiments(options.getExperiments());
    // Set the Docker container image that executes Dataflow worker harness, residing in Google
    // Container Registry. Translator is guaranteed to create a worker pool prior to this point.
    String workerHarnessContainerImage = getContainerImageForJob(options);
    for (WorkerPool workerPool : newJob.getEnvironment().getWorkerPools()) {
        workerPool.setWorkerHarnessContainerImage(workerHarnessContainerImage);
    }
    newJob.getEnvironment().setVersion(getEnvironmentVersion(options));
    if (hooks != null) {
        hooks.modifyEnvironmentBeforeSubmission(newJob.getEnvironment());
    }
    if (!isNullOrEmpty(options.getDataflowJobFile()) || !isNullOrEmpty(options.getTemplateLocation())) {
        boolean isTemplate = !isNullOrEmpty(options.getTemplateLocation());
        if (isTemplate) {
            checkArgument(isNullOrEmpty(options.getDataflowJobFile()), "--dataflowJobFile and --templateLocation are mutually exclusive.");
        }
        String fileLocation = firstNonNull(options.getTemplateLocation(), options.getDataflowJobFile());
        checkArgument(fileLocation.startsWith("/") || fileLocation.startsWith("gs://"), "Location must be local or on Cloud Storage, got %s.", fileLocation);
        ResourceId fileResource = FileSystems.matchNewResource(fileLocation, false);
        String workSpecJson = DataflowPipelineTranslator.jobToString(newJob);
        try (PrintWriter printWriter = new PrintWriter(Channels.newOutputStream(FileSystems.create(fileResource, MimeTypes.TEXT)))) {
            printWriter.print(workSpecJson);
            LOG.info("Printed job specification to {}", fileLocation);
        } catch (IOException ex) {
            String error = String.format("Cannot create output file at %s", fileLocation);
            if (isTemplate) {
                throw new RuntimeException(error, ex);
            } else {
                LOG.warn(error, ex);
            }
        }
        if (isTemplate) {
            LOG.info("Template successfully created.");
            return new DataflowTemplateJob();
        }
    }
    String jobIdToUpdate = null;
    if (options.isUpdate()) {
        jobIdToUpdate = getJobIdFromName(options.getJobName());
        newJob.setTransformNameMapping(options.getTransformNameMapping());
        newJob.setReplaceJobId(jobIdToUpdate);
    }
    Job jobResult;
    try {
        jobResult = dataflowClient.createJob(newJob);
    } catch (GoogleJsonResponseException e) {
        String errorMessages = "Unexpected errors";
        if (e.getDetails() != null) {
            if (Utf8.encodedLength(newJob.toString()) >= CREATE_JOB_REQUEST_LIMIT_BYTES) {
                errorMessages = "The size of the serialized JSON representation of the pipeline " + "exceeds the allowable limit. " + "For more information, please check the FAQ link below:\n" + "https://cloud.google.com/dataflow/faq";
            } else {
                errorMessages = e.getDetails().getMessage();
            }
        }
        throw new RuntimeException("Failed to create a workflow job: " + errorMessages, e);
    } catch (IOException e) {
        throw new RuntimeException("Failed to create a workflow job", e);
    }
    // Use a raw client for post-launch monitoring, as status calls may fail
    // regularly and need not be retried automatically.
    DataflowPipelineJob dataflowPipelineJob = new DataflowPipelineJob(DataflowClient.create(options), jobResult.getId(), options, jobSpecification.getStepNames());
    // depending on whether this is a reload or not.
    if (jobResult.getClientRequestId() != null && !jobResult.getClientRequestId().isEmpty() && !jobResult.getClientRequestId().equals(requestId)) {
        // If updating a job.
        if (options.isUpdate()) {
            throw new DataflowJobAlreadyUpdatedException(dataflowPipelineJob, String.format("The job named %s with id: %s has already been updated into job id: %s " + "and cannot be updated again.", newJob.getName(), jobIdToUpdate, jobResult.getId()));
        } else {
            throw new DataflowJobAlreadyExistsException(dataflowPipelineJob, String.format("There is already an active job named %s with id: %s. If you want " + "to submit a second job, try again by setting a different name using --jobName.", newJob.getName(), jobResult.getId()));
        }
    }
    LOG.info("To access the Dataflow monitoring console, please navigate to {}", MonitoringUtil.getJobMonitoringPageURL(options.getProject(), jobResult.getId()));
    System.out.println("Submitted job: " + jobResult.getId());
    LOG.info("To cancel the job using the 'gcloud' tool, run:\n> {}", MonitoringUtil.getGcloudCancelCommand(options, jobResult.getId()));
    return dataflowPipelineJob;
}