Examples with DataxProcessor com.qlangtech.tis.datax.impl.DataxProcessor used on opensource projects

Example 16 with DataxProcessor

use of com.qlangtech.tis.datax.impl.DataxProcessor in project tis by qlangtech.

the class DataXExecuteInterceptor method execute.

@Override
protected ExecuteResult execute(IExecChainContext execChainContext) throws Exception {
    int nThreads = 2;
    final ExecutorService executorService = new ThreadPoolExecutor(nThreads, nThreads, 0L, TimeUnit.MILLISECONDS, new LinkedBlockingQueue<>(DataXJobSubmit.MAX_TABS_NUM_IN_PER_JOB), Executors.defaultThreadFactory());
    final Map<String, TISReactor.TaskAndMilestone> taskMap = Maps.newHashMap();
    RpcServiceReference statusRpc = getDataXExecReporter();
    DataxProcessor appSource = execChainContext.getAppSource();
    IRemoteTaskTrigger jobTrigger = null;
    RunningStatus runningStatus = null;
    List<IRemoteTaskTrigger> triggers = Lists.newArrayList();
    List<File> cfgFileNames = appSource.getDataxCfgFileNames(null);
    if (CollectionUtils.isEmpty(cfgFileNames)) {
        throw new IllegalStateException("dataX cfgFileNames can not be empty");
    }
    DataXJobSubmit.InstanceType expectDataXJobSumit = getDataXTriggerType();
    Optional<DataXJobSubmit> jobSubmit = DataXJobSubmit.getDataXJobSubmit(expectDataXJobSumit);
    // 如果分布式worker ready的话
    if (!jobSubmit.isPresent()) {
        throw new IllegalStateException("can not find expect jobSubmit by type:" + expectDataXJobSumit);
    }
    DataXJobSubmit submit = jobSubmit.get();
    final DataXJobSubmit.IDataXJobContext dataXJobContext = submit.createJobContext(execChainContext);
    Objects.requireNonNull(dataXJobContext, "dataXJobContext can not be null");
    try {
        DumpPhaseStatus dumpStatus = this.getPhaseStatus(execChainContext, FullbuildPhase.FullDump);
        for (File fileName : cfgFileNames) {
            jobTrigger = createDataXJob(dataXJobContext, submit, expectDataXJobSumit, statusRpc, appSource, fileName.getName());
            triggers.add(jobTrigger);
            taskMap.put(fileName.getName(), new TISReactor.TaskAndMilestone(DataflowTask.createDumpTask(jobTrigger)));
            // StatusRpcClient.AssembleSvcCompsite svc = statusRpc.get();
            // 将任务注册，可供页面展示
            // svc.reportDumpJobStatus(false, false, true, execChainContext.getTaskId()
            // , fileName.getName(), 0, 0);
            dumpStatus.getTable(fileName.getName()).setWaiting(true);
        }
        logger.info("trigger dataX jobs by mode:{},with:{}", this.getDataXTriggerType(), cfgFileNames.stream().map((f) -> f.getName()).collect(Collectors.joining(",")));
        // for (IRemoteJobTrigger t : triggers) {
        // t.submitJob();
        // }
        IDataxReader reader = appSource.getReader(null);
        List<ISelectedTab> selectedTabs = reader.getSelectedTabs();
        // Map<String, IDataxProcessor.TableAlias> tabAlias = appSource.getTabAlias();
        IDataxWriter writer = appSource.getWriter(null);
        if (writer instanceof IDataXBatchPost) {
            IDataXBatchPost batchPostTask = (IDataXBatchPost) writer;
            JoinPhaseStatus phaseStatus = this.getPhaseStatus(execChainContext, FullbuildPhase.JOIN);
            for (ISelectedTab entry : selectedTabs) {
                IRemoteTaskTrigger postTaskTrigger = batchPostTask.createPostTask(execChainContext, entry);
                triggers.add(postTaskTrigger);
                JoinPhaseStatus.JoinTaskStatus taskStatus = phaseStatus.getTaskStatus(postTaskTrigger.getTaskName());
                taskStatus.setWaiting(true);
                taskMap.put(postTaskTrigger.getTaskName(), new TISReactor.TaskAndMilestone(createJoinTask(postTaskTrigger, taskStatus)));
            }
        }
        // example: "->a ->b a,b->c"
        String dagSessionSpec = triggers.stream().map((trigger) -> {
            List<String> dpts = trigger.getTaskDependencies();
            return dpts.stream().collect(Collectors.joining(",")) + "->" + trigger.getTaskName();
        }).collect(Collectors.joining(" "));
        logger.info("dataX:{} of dagSessionSpec:{}", execChainContext.getIndexName(), dagSessionSpec);
        ExecuteResult[] faildResult = new ExecuteResult[] { ExecuteResult.createSuccess() };
        this.executeDAG(executorService, execChainContext, dagSessionSpec, taskMap, new ReactorListener() {

            @Override
            public void onTaskCompleted(Task t) {
            // dumpPhaseStatus.isComplete();
            // joinPhaseStatus.isComplete();
            }

            @Override
            public void onTaskFailed(Task t, Throwable err, boolean fatal) {
                logger.error(t.getDisplayName(), err);
                faildResult[0] = ExecuteResult.createFaild().setMessage("status.runningStatus.isComplete():" + err.getMessage());
                if (err instanceof InterruptedException) {
                    logger.warn("DataX Name:{},taskid:{} has been canceled", execChainContext.getIndexName(), execChainContext.getTaskId());
                    // this job has been cancel, trigger from TisServlet.doDelete()
                    for (IRemoteTaskTrigger tt : triggers) {
                        try {
                            tt.cancel();
                        } catch (Throwable ex) {
                        }
                    }
                }
            }
        });
        // ExecuteResult result = new ExecuteResult(!faild);
        for (IRemoteTaskTrigger trigger : triggers) {
            if (trigger.isAsyn()) {
                execChainContext.addAsynSubJob(new IExecChainContext.AsynSubJob(trigger.getAsynJobName()));
            }
        }
        return faildResult[0];
    } finally {
        try {
            dataXJobContext.destroy();
        } catch (Throwable e) {
            logger.error(e.getMessage(), e);
        }
    }
}

Example 17 with DataxProcessor

use of com.qlangtech.tis.datax.impl.DataxProcessor in project tis by qlangtech.

the class DataxExecutor method exec.

public void exec(Integer jobId, String jobName, String dataxName, String execTimeStamp) throws Exception {
    boolean success = false;
    MDC.put(IParamContext.KEY_TASK_ID, String.valueOf(jobId));
    try {
        logger.info("process DataX job, dataXName:{},jobid:{},jobName:{}", dataxName, jobId, jobName);
        DataxExecutor.synchronizeDataXPluginsFromRemoteRepository(dataxName, jobName);
        final JarLoader uberClassLoader = new TISJarLoader(TIS.get().getPluginManager());
        DataxProcessor dataxProcessor = DataxProcessor.load(null, dataxName);
        this.startWork(dataxName, jobId, jobName, execTimeStamp, dataxProcessor, uberClassLoader);
        success = true;
    } finally {
        TIS.clean();
        if (execMode == DataXJobSubmit.InstanceType.DISTRIBUTE) {
            try {
                DagTaskUtils.feedbackAsynTaskStatus(jobId, jobName, success);
            } catch (Throwable e) {
                logger.warn("notify exec result faild,jobId:" + jobId + ",jobName:" + jobName, e);
            }
        }
    }
}

Example 18 with DataxProcessor

use of com.qlangtech.tis.datax.impl.DataxProcessor in project tis by qlangtech.

the class TestStreamComponentCodeGeneratorFlink method testSingleTableCodeGenerator.

/**
 * 测试单表增量脚本生成
 *
 * @throws Exception
 */
public void testSingleTableCodeGenerator() throws Exception {
    // CoreAction.create
    // String topologyName = "employees4local";
    String collectionName = "mysql_elastic";
    IAppSource appSource = IAppSource.load(null, collectionName);
    assertTrue(appSource instanceof DataxProcessor);
    DataxProcessor dataXProcessor = (DataxProcessor) appSource;
    // Optional<ERRules> erRule = ERRules.getErRule(topologyName);
    // // 测试针对单表的的topology增量脚本生成
    long timestamp = 20191111115959l;
    // SqlTaskNodeMeta.SqlDataFlowTopology topology = SqlTaskNodeMeta.getSqlDataFlowTopology(topologyName);
    // assertNotNull(topology);
    // if (!erRule.isPresent()) {
    // ERRules.createDefaultErRule(topology);
    // }
    List<FacadeContext> facadeList = Lists.newArrayList();
    StreamComponentCodeGeneratorFlink streamCodeGenerator = new StreamComponentCodeGeneratorFlink(collectionName, timestamp, facadeList, (IBasicAppSource) appSource);
    // EasyMock.replay(streamIncrGenerateStrategy);
    streamCodeGenerator.build();
    TestStreamComponentCodeGenerator.assertGenerateContentEqual(timestamp, collectionName, "MysqlElasticListener.scala");
// EasyMock.verify(streamIncrGenerateStrategy);
}