Examples with DataStagingTaskModel org.apache.airavata.model.task.DataStagingTaskModel used on opensource projects

Example 1 with DataStagingTaskModel

use of org.apache.airavata.model.task.DataStagingTaskModel in project airavata by apache.

the class GFacEngineImpl method inputDataStaging.

private boolean inputDataStaging(TaskContext taskContext, boolean recover) throws GFacException, TException {
    TaskStatus taskStatus = new TaskStatus(TaskState.EXECUTING);
    taskStatus.setTimeOfStateChange(AiravataUtils.getCurrentTimestamp().getTime());
    taskContext.setTaskStatus(taskStatus);
    GFacUtils.saveAndPublishTaskStatus(taskContext);
    ProcessContext processContext = taskContext.getParentProcessContext();
    // handle URI_COLLECTION input data type
    Task dMoveTask = Factory.getDataMovementTask(processContext.getDataMovementProtocol());
    if (null == dMoveTask) {
        throw new GFacException("Unsupported security protocol, Airavata doesn't support " + processContext.getDataMovementProtocol() + " protocol yet.");
    }
    if (taskContext.getProcessInput().getType() == DataType.URI_COLLECTION) {
        String values = taskContext.getProcessInput().getValue();
        String[] multiple_inputs = values.split(GFacConstants.MULTIPLE_INPUTS_SPLITTER);
        DataStagingTaskModel subTaskModel = (DataStagingTaskModel) taskContext.getSubTaskModel();
        for (String input : multiple_inputs) {
            taskContext.getProcessInput().setValue(input);
            subTaskModel.setSource(input);
            taskStatus = executeTask(taskContext, dMoveTask, false);
        }
        taskContext.getProcessInput().setValue(values);
    } else {
        taskStatus = executeTask(taskContext, dMoveTask, false);
    }
    taskStatus.setTimeOfStateChange(AiravataUtils.getCurrentTimestamp().getTime());
    taskContext.setTaskStatus(taskStatus);
    GFacUtils.saveAndPublishTaskStatus(taskContext);
    checkFailures(taskContext, taskStatus, dMoveTask);
    return false;
}

Example 2 with DataStagingTaskModel

use of org.apache.airavata.model.task.DataStagingTaskModel in project airavata by apache.

the class GFacEngineImpl method executeTaskListFrom.

private void executeTaskListFrom(ProcessContext processContext, String startingTaskId) throws GFacException {
    // checkpoint
    if (processContext.isInterrupted() && processContext.getProcessState() != ProcessState.MONITORING) {
        GFacUtils.handleProcessInterrupt(processContext);
        return;
    }
    List<TaskModel> taskList = processContext.getTaskList();
    Map<String, TaskModel> taskMap = processContext.getTaskMap();
    boolean fastForward = true;
    for (String taskId : processContext.getTaskExecutionOrder()) {
        if (fastForward) {
            if (taskId.equalsIgnoreCase(startingTaskId)) {
                fastForward = false;
            } else {
                continue;
            }
        }
        TaskModel taskModel = taskMap.get(taskId);
        processContext.setCurrentExecutingTaskModel(taskModel);
        TaskTypes taskType = taskModel.getTaskType();
        TaskContext taskContext = getTaskContext(processContext);
        taskContext.setTaskModel(taskModel);
        ProcessStatus status = null;
        switch(taskType) {
            case ENV_SETUP:
                status = new ProcessStatus(ProcessState.CONFIGURING_WORKSPACE);
                status.setTimeOfStateChange(AiravataUtils.getCurrentTimestamp().getTime());
                processContext.setProcessStatus(status);
                GFacUtils.saveAndPublishProcessStatus(processContext);
                // checkpoint
                if (processContext.isInterrupted()) {
                    GFacUtils.handleProcessInterrupt(processContext);
                    return;
                }
                configureWorkspace(taskContext, processContext.isRecovery());
                // checkpoint
                if (processContext.isInterrupted()) {
                    GFacUtils.handleProcessInterrupt(processContext);
                    return;
                }
                break;
            case DATA_STAGING:
                try {
                    // checkpoint
                    if (processContext.isInterrupted()) {
                        GFacUtils.handleProcessInterrupt(processContext);
                        return;
                    }
                    DataStagingTaskModel subTaskModel = (DataStagingTaskModel) taskContext.getSubTaskModel();
                    DataStageType type = subTaskModel.getType();
                    switch(type) {
                        case INPUT:
                            status = new ProcessStatus(ProcessState.INPUT_DATA_STAGING);
                            status.setTimeOfStateChange(AiravataUtils.getCurrentTimestamp().getTime());
                            processContext.setProcessStatus(status);
                            GFacUtils.saveAndPublishProcessStatus(processContext);
                            taskContext.setProcessInput(subTaskModel.getProcessInput());
                            inputDataStaging(taskContext, processContext.isRecovery());
                            break;
                        case OUPUT:
                            status = new ProcessStatus(ProcessState.OUTPUT_DATA_STAGING);
                            status.setTimeOfStateChange(AiravataUtils.getCurrentTimestamp().getTime());
                            processContext.setProcessStatus(status);
                            GFacUtils.saveAndPublishProcessStatus(processContext);
                            taskContext.setProcessOutput(subTaskModel.getProcessOutput());
                            outputDataStaging(taskContext, processContext.isRecovery(), false);
                            break;
                        case ARCHIVE_OUTPUT:
                            status = new ProcessStatus(ProcessState.OUTPUT_DATA_STAGING);
                            status.setTimeOfStateChange(AiravataUtils.getCurrentTimestamp().getTime());
                            processContext.setProcessStatus(status);
                            GFacUtils.saveAndPublishProcessStatus(processContext);
                            outputDataStaging(taskContext, processContext.isRecovery(), true);
                            break;
                    }
                    // checkpoint
                    if (processContext.isInterrupted()) {
                        GFacUtils.handleProcessInterrupt(processContext);
                        return;
                    }
                } catch (TException e) {
                    throw new GFacException(e);
                }
                break;
            case JOB_SUBMISSION:
                // checkpoint
                if (processContext.isInterrupted()) {
                    GFacUtils.handleProcessInterrupt(processContext);
                    return;
                }
                status = new ProcessStatus(ProcessState.EXECUTING);
                status.setTimeOfStateChange(AiravataUtils.getCurrentTimestamp().getTime());
                processContext.setProcessStatus(status);
                GFacUtils.saveAndPublishProcessStatus(processContext);
                executeJobSubmission(taskContext, processContext.isRecovery());
                // Don't put any checkpoint in between JobSubmission and Monitoring tasks
                JobStatus jobStatus = processContext.getJobModel().getJobStatuses().get(0);
                if (jobStatus != null && (jobStatus.getJobState() == JobState.SUBMITTED || jobStatus.getJobState() == JobState.QUEUED || jobStatus.getJobState() == JobState.ACTIVE)) {
                    List<OutputDataObjectType> processOutputs = processContext.getProcessModel().getProcessOutputs();
                    if (processOutputs != null && !processOutputs.isEmpty()) {
                        for (OutputDataObjectType output : processOutputs) {
                            try {
                                if (output.isOutputStreaming()) {
                                    TaskModel streamingTaskModel = new TaskModel();
                                    streamingTaskModel.setTaskType(TaskTypes.OUTPUT_FETCHING);
                                    streamingTaskModel.setTaskStatuses(Arrays.asList(new TaskStatus(TaskState.CREATED)));
                                    streamingTaskModel.setCreationTime(AiravataUtils.getCurrentTimestamp().getTime());
                                    streamingTaskModel.setParentProcessId(processContext.getProcessId());
                                    TaskContext streamingTaskContext = getTaskContext(processContext);
                                    DataStagingTaskModel submodel = new DataStagingTaskModel();
                                    submodel.setType(DataStageType.OUPUT);
                                    submodel.setProcessOutput(output);
                                    URI source = new URI(processContext.getDataMovementProtocol().name(), processContext.getComputeResourceLoginUserName(), processContext.getComputeResourceDescription().getHostName(), 22, processContext.getWorkingDir() + output.getValue(), null, null);
                                    submodel.setSource(source.getPath());
                                    submodel.setDestination("dummy://temp/file/location");
                                    streamingTaskModel.setSubTaskModel(ThriftUtils.serializeThriftObject(submodel));
                                    String streamTaskId = (String) processContext.getExperimentCatalog().add(ExpCatChildDataType.TASK, streamingTaskModel, processContext.getProcessId());
                                    streamingTaskModel.setTaskId(streamTaskId);
                                    streamingTaskContext.setTaskModel(streamingTaskModel);
                                    executeDataStreaming(streamingTaskContext, processContext.isRecovery());
                                }
                            } catch (URISyntaxException | TException | RegistryException e) {
                                log.error("Error while streaming output " + output.getValue());
                            }
                        }
                    }
                }
                break;
            case MONITORING:
                status = new ProcessStatus(ProcessState.MONITORING);
                status.setTimeOfStateChange(AiravataUtils.getCurrentTimestamp().getTime());
                processContext.setProcessStatus(status);
                GFacUtils.saveAndPublishProcessStatus(processContext);
                executeJobMonitoring(taskContext, processContext.isRecovery());
                break;
            case ENV_CLEANUP:
                // TODO implement environment clean up task logic
                break;
            default:
                throw new GFacException("Unsupported Task type");
        }
        if (processContext.isPauseTaskExecution()) {
            // If any task put processContext to wait, the same task must continue processContext execution.
            return;
        }
    }
    processContext.setComplete(true);
}

Example 3 with DataStagingTaskModel

use of org.apache.airavata.model.task.DataStagingTaskModel in project airavata by apache.

the class DataStageTask method execute.

@Override
public TaskStatus execute(TaskContext taskContext) {
    TaskStatus status = new TaskStatus(TaskState.COMPLETED);
    if (taskContext.getTaskModel().getTaskType() != TaskTypes.DATA_STAGING) {
        status.setState(TaskState.FAILED);
        status.setReason("Invalid task call, expected " + TaskTypes.DATA_STAGING.toString() + " but found " + taskContext.getTaskModel().getTaskType().toString());
    } else {
        try {
            DataStagingTaskModel subTaskModel = ((DataStagingTaskModel) taskContext.getSubTaskModel());
            URI sourceURI = new URI(subTaskModel.getSource());
            URI destinationURI = new URI(subTaskModel.getDestination());
            ProcessState processState = taskContext.getParentProcessContext().getProcessState();
            if (processState == ProcessState.INPUT_DATA_STAGING) {
                /**
                 * copy local file to compute resource.
                 */
                taskContext.getParentProcessContext().getDataMovementRemoteCluster().copyTo(sourceURI.getPath(), destinationURI.getPath());
            } else if (processState == ProcessState.OUTPUT_DATA_STAGING) {
                /**
                 * copy remote file from compute resource.
                 */
                taskContext.getParentProcessContext().getDataMovementRemoteCluster().copyFrom(sourceURI.getPath(), destinationURI.getPath());
            }
            status.setReason("Successfully staged data");
        } catch (GFacException e) {
            String msg = "Scp attempt failed";
            log.error(msg, e);
            status.setState(TaskState.FAILED);
            status.setReason(msg);
            ErrorModel errorModel = new ErrorModel();
            errorModel.setActualErrorMessage(e.getMessage());
            errorModel.setUserFriendlyMessage(msg);
            taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
        } catch (TException e) {
            String msg = "Invalid task invocation";
            log.error(msg, e);
            status.setState(TaskState.FAILED);
            status.setReason(msg);
            ErrorModel errorModel = new ErrorModel();
            errorModel.setActualErrorMessage(e.getMessage());
            errorModel.setUserFriendlyMessage(msg);
            taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
        } catch (URISyntaxException e) {
            String msg = "source or destination is not a valid URI";
            log.error(msg, e);
            status.setState(TaskState.FAILED);
            status.setReason(msg);
            ErrorModel errorModel = new ErrorModel();
            errorModel.setActualErrorMessage(e.getMessage());
            errorModel.setUserFriendlyMessage(msg);
            taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
        }
    }
    return status;
}

Example 4 with DataStagingTaskModel

use of org.apache.airavata.model.task.DataStagingTaskModel in project airavata by apache.

the class DataStreamingTask method execute.

@Override
public TaskStatus execute(TaskContext taskContext) {
    ProcessState processState = taskContext.getParentProcessContext().getProcessState();
    try {
        TaskStatus status = new TaskStatus(TaskState.EXECUTING);
        final DataStagingTaskModel subTaskModel = (DataStagingTaskModel) ThriftUtils.getSubTaskModel(taskContext.getTaskModel());
        if (processState == ProcessState.OUTPUT_DATA_STAGING) {
            OutputDataObjectType processOutput = taskContext.getProcessOutput();
            if (processOutput != null && processOutput.getValue() == null) {
                log.error("expId: {}, processId:{}, taskId: {}:- Couldn't stage file {} , file name shouldn't be null", taskContext.getExperimentId(), taskContext.getProcessId(), taskContext.getTaskId(), processOutput.getName());
                status = new TaskStatus(TaskState.FAILED);
                if (processOutput.isIsRequired()) {
                    status.setReason("File name is null, but this output's isRequired bit is not set");
                } else {
                    status.setReason("File name is null");
                }
                return status;
            }
            if (processOutput != null) {
                if (processOutput.isOutputStreaming()) {
                    // stream output periodically
                    ComputationalResourceSchedulingModel resourceSchedule = taskContext.getParentProcessContext().getProcessModel().getProcessResourceSchedule();
                    int wallTimeLimit = resourceSchedule.getWallTimeLimit();
                    if (wallTimeLimit > 10) {
                        int period = wallTimeLimit / 10;
                        Timer timer = new Timer();
                        StreamData streamData = new StreamData(userName, hostName, inputPath, taskContext, subTaskModel);
                        timer.schedule(streamData, 0, 1000 * 60 * period);
                        status.setState(TaskState.COMPLETED);
                    }
                }
            }
        }
        return null;
    } catch (TException e) {
        log.error("Error while creating data streaming task", e);
        return null;
    }
}

Example 5 with DataStagingTaskModel

use of org.apache.airavata.model.task.DataStagingTaskModel in project airavata by apache.

the class SCPDataStageTask method execute.

@Override
public TaskStatus execute(TaskContext taskContext) {
    TaskStatus status = new TaskStatus(TaskState.EXECUTING);
    AuthenticationInfo authenticationInfo = null;
    DataStagingTaskModel subTaskModel = null;
    String localDataDir = null;
    ProcessContext processContext = taskContext.getParentProcessContext();
    ProcessState processState = processContext.getProcessState();
    try {
        subTaskModel = ((DataStagingTaskModel) taskContext.getSubTaskModel());
        if (processState == ProcessState.OUTPUT_DATA_STAGING) {
            OutputDataObjectType processOutput = taskContext.getProcessOutput();
            if (processOutput != null && processOutput.getValue() == null) {
                log.error("expId: {}, processId:{}, taskId: {}:- Couldn't stage file {} , file name shouldn't be null", taskContext.getExperimentId(), taskContext.getProcessId(), taskContext.getTaskId(), processOutput.getName());
                status = new TaskStatus(TaskState.FAILED);
                if (processOutput.isIsRequired()) {
                    status.setReason("File name is null, but this output's isRequired bit is not set");
                } else {
                    status.setReason("File name is null");
                }
                return status;
            }
        } else if (processState == ProcessState.INPUT_DATA_STAGING) {
            InputDataObjectType processInput = taskContext.getProcessInput();
            if (processInput != null && processInput.getValue() == null) {
                log.error("expId: {}, processId:{}, taskId: {}:- Couldn't stage file {} , file name shouldn't be null", taskContext.getExperimentId(), taskContext.getProcessId(), taskContext.getTaskId(), processInput.getName());
                status = new TaskStatus(TaskState.FAILED);
                if (processInput.isIsRequired()) {
                    status.setReason("File name is null, but this input's isRequired bit is not set");
                } else {
                    status.setReason("File name is null");
                }
                return status;
            }
        } else {
            status.setState(TaskState.FAILED);
            status.setReason("Invalid task invocation, Support " + ProcessState.INPUT_DATA_STAGING.name() + " and " + "" + ProcessState.OUTPUT_DATA_STAGING.name() + " process phases. found " + processState.name());
            return status;
        }
        StorageResourceDescription storageResource = processContext.getStorageResource();
        // StoragePreference storagePreference = taskContext.getParentProcessContext().getStoragePreference();
        String hostName = null;
        if (storageResource != null) {
            hostName = storageResource.getHostName();
        } else {
            throw new GFacException("Storage Resource is null");
        }
        String inputPath = processContext.getStorageFileSystemRootLocation();
        inputPath = (inputPath.endsWith(File.separator) ? inputPath : inputPath + File.separator);
        // use rsync instead of scp if source and destination host and user name is same.
        URI sourceURI = new URI(subTaskModel.getSource());
        String fileName = sourceURI.getPath().substring(sourceURI.getPath().lastIndexOf(File.separator) + 1, sourceURI.getPath().length());
        Session remoteSession = Factory.getSSHSession(Factory.getComputerResourceSSHKeyAuthentication(processContext), processContext.getComputeResourceServerInfo());
        Session storageSession = Factory.getSSHSession(Factory.getStorageSSHKeyAuthentication(processContext), processContext.getStorageResourceServerInfo());
        URI destinationURI = null;
        if (subTaskModel.getDestination().startsWith("dummy")) {
            destinationURI = TaskUtils.getDestinationURI(taskContext, hostName, inputPath, fileName);
            subTaskModel.setDestination(destinationURI.toString());
        } else {
            destinationURI = new URI(subTaskModel.getDestination());
        }
        if (sourceURI.getHost().equalsIgnoreCase(destinationURI.getHost()) && sourceURI.getUserInfo().equalsIgnoreCase(destinationURI.getUserInfo())) {
            localDataCopy(taskContext, sourceURI, destinationURI);
            status.setState(TaskState.COMPLETED);
            status.setReason("Locally copied file using 'cp' command ");
            return status;
        }
        status = new TaskStatus(TaskState.COMPLETED);
        // Wildcard for file name. Has to find the correct name.
        if (fileName.contains("*")) {
            String destParentPath = (new File(destinationURI.getPath())).getParentFile().getPath();
            String sourceParentPath = (new File(sourceURI.getPath())).getParentFile().getPath();
            List<String> fileNames = taskContext.getParentProcessContext().getDataMovementRemoteCluster().getFileNameFromExtension(fileName, sourceParentPath, remoteSession);
            ExperimentCatalog experimentCatalog = processContext.getExperimentCatalog();
            String experimentId = processContext.getExperimentId();
            String processId = processContext.getProcessId();
            OutputDataObjectType processOutput = taskContext.getProcessOutput();
            for (int i = 0; i < fileNames.size(); i++) {
                String temp = fileNames.get(i);
                if (temp != null && temp != "") {
                    fileName = temp;
                }
                if (destParentPath.endsWith(File.separator)) {
                    destinationURI = new URI(destParentPath + fileName);
                } else {
                    destinationURI = new URI(destParentPath + File.separator + fileName);
                }
                // Wildcard support is only enabled for output data staging
                if (processState == ProcessState.OUTPUT_DATA_STAGING) {
                    processOutput.setName(fileName);
                    experimentCatalog.add(ExpCatChildDataType.EXPERIMENT_OUTPUT, Arrays.asList(processOutput), experimentId);
                    experimentCatalog.add(ExpCatChildDataType.PROCESS_OUTPUT, Arrays.asList(processOutput), processId);
                    taskContext.setProcessOutput(processOutput);
                    makeDir(taskContext, destinationURI);
                    // TODO - save updated subtask model with new destination
                    outputDataStaging(taskContext, remoteSession, sourceURI, storageSession, destinationURI);
                    status.setReason("Successfully staged output data");
                }
            }
            if (processState == ProcessState.OUTPUT_DATA_STAGING) {
                status.setReason("Successfully staged output data");
            } else {
                status.setReason("Wildcard support is only enabled for output data staging");
            }
        } else {
            if (processState == ProcessState.INPUT_DATA_STAGING) {
                inputDataStaging(taskContext, storageSession, sourceURI, remoteSession, destinationURI);
                status.setReason("Successfully staged input data");
            } else if (processState == ProcessState.OUTPUT_DATA_STAGING) {
                makeDir(taskContext, destinationURI);
                // TODO - save updated subtask model with new destination
                outputDataStaging(taskContext, remoteSession, sourceURI, storageSession, destinationURI);
                status.setReason("Successfully staged output data");
            }
        }
    } catch (TException e) {
        String msg = "Couldn't create subTask model thrift model";
        log.error(msg, e);
        status.setState(TaskState.FAILED);
        status.setReason(msg);
        ErrorModel errorModel = new ErrorModel();
        errorModel.setActualErrorMessage(e.getMessage());
        errorModel.setUserFriendlyMessage(msg);
        taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
        return status;
    } catch (ApplicationSettingsException | FileNotFoundException e) {
        String msg = "Failed while reading credentials";
        log.error(msg, e);
        status.setState(TaskState.FAILED);
        status.setReason(msg);
        ErrorModel errorModel = new ErrorModel();
        errorModel.setActualErrorMessage(e.getMessage());
        errorModel.setUserFriendlyMessage(msg);
        taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
    } catch (URISyntaxException e) {
        String msg = "Source or destination uri is not correct source : " + subTaskModel.getSource() + ", " + "destination : " + subTaskModel.getDestination();
        log.error(msg, e);
        status.setState(TaskState.FAILED);
        status.setReason(msg);
        ErrorModel errorModel = new ErrorModel();
        errorModel.setActualErrorMessage(e.getMessage());
        errorModel.setUserFriendlyMessage(msg);
        taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
    } catch (CredentialStoreException e) {
        String msg = "Storage authentication issue, could be invalid credential token";
        log.error(msg, e);
        status.setState(TaskState.FAILED);
        status.setReason(msg);
        ErrorModel errorModel = new ErrorModel();
        errorModel.setActualErrorMessage(e.getMessage());
        errorModel.setUserFriendlyMessage(msg);
        taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
    } catch (AiravataException e) {
        String msg = "Error while creating ssh session with client";
        log.error(msg, e);
        status.setState(TaskState.FAILED);
        status.setReason(msg);
        ErrorModel errorModel = new ErrorModel();
        errorModel.setActualErrorMessage(e.getMessage());
        errorModel.setUserFriendlyMessage(msg);
        taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
    } catch (JSchException | IOException e) {
        String msg = "Failed to do scp with client";
        log.error(msg, e);
        status.setState(TaskState.FAILED);
        status.setReason(msg);
        ErrorModel errorModel = new ErrorModel();
        errorModel.setActualErrorMessage(e.getMessage());
        errorModel.setUserFriendlyMessage(msg);
        taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
    } catch (RegistryException | GFacException e) {
        String msg = "Data staging failed";
        log.error(msg, e);
        status.setState(TaskState.FAILED);
        status.setReason(msg);
        ErrorModel errorModel = new ErrorModel();
        errorModel.setActualErrorMessage(e.getMessage());
        errorModel.setUserFriendlyMessage(msg);
        taskContext.getTaskModel().setTaskErrors(Arrays.asList(errorModel));
    }
    return status;
}