Search in sources :

Example 1 with IDataXBatchPost

use of com.qlangtech.tis.datax.IDataXBatchPost 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);
        }
    }
}
Also used : ExecuteResult(com.qlangtech.tis.exec.ExecuteResult) java.util(java.util) IDataxWriter(com.qlangtech.tis.datax.IDataxWriter) IExecChainContext(com.qlangtech.tis.exec.IExecChainContext) ISelectedTab(com.qlangtech.tis.plugin.ds.ISelectedTab) LoggerFactory(org.slf4j.LoggerFactory) AtomicReference(java.util.concurrent.atomic.AtomicReference) RpcServiceReference(com.tis.hadoop.rpc.RpcServiceReference) Lists(com.google.common.collect.Lists) DataxProcessor(com.qlangtech.tis.datax.impl.DataxProcessor) CollectionUtils(org.apache.commons.collections.CollectionUtils) DataXJobSubmit(com.qlangtech.tis.datax.DataXJobSubmit) JoinPhaseStatus(com.qlangtech.tis.fullbuild.phasestatus.impl.JoinPhaseStatus) IDataxReader(com.qlangtech.tis.datax.IDataxReader) AdapterStatusUmbilicalProtocol(com.qlangtech.tis.realtime.yarn.rpc.impl.AdapterStatusUmbilicalProtocol) DumpPhaseStatus(com.qlangtech.tis.fullbuild.phasestatus.impl.DumpPhaseStatus) Task(org.jvnet.hudson.reactor.Task) IDataXBatchPost(com.qlangtech.tis.datax.IDataXBatchPost) TrackableExecuteInterceptor(com.qlangtech.tis.exec.impl.TrackableExecuteInterceptor) FullbuildPhase(com.qlangtech.tis.assemble.FullbuildPhase) IncrStatusUmbilicalProtocolImpl(com.qlangtech.tis.rpc.server.IncrStatusUmbilicalProtocolImpl) IRemoteTaskTrigger(com.qlangtech.tis.fullbuild.indexbuild.IRemoteTaskTrigger) Logger(org.slf4j.Logger) java.util.concurrent(java.util.concurrent) IncrStatusUmbilicalProtocol(com.qlangtech.tis.realtime.yarn.rpc.IncrStatusUmbilicalProtocol) ReactorListener(org.jvnet.hudson.reactor.ReactorListener) Maps(com.google.common.collect.Maps) Collectors(java.util.stream.Collectors) File(java.io.File) RunningStatus(com.qlangtech.tis.fullbuild.indexbuild.RunningStatus) DataflowTask(com.qlangtech.tis.fullbuild.taskflow.DataflowTask) TISReactor(com.qlangtech.tis.fullbuild.taskflow.TISReactor) ITISRpcService(com.tis.hadoop.rpc.ITISRpcService) Task(org.jvnet.hudson.reactor.Task) DataflowTask(com.qlangtech.tis.fullbuild.taskflow.DataflowTask) DataXJobSubmit(com.qlangtech.tis.datax.DataXJobSubmit) ISelectedTab(com.qlangtech.tis.plugin.ds.ISelectedTab) RpcServiceReference(com.tis.hadoop.rpc.RpcServiceReference) DumpPhaseStatus(com.qlangtech.tis.fullbuild.phasestatus.impl.DumpPhaseStatus) IDataXBatchPost(com.qlangtech.tis.datax.IDataXBatchPost) IExecChainContext(com.qlangtech.tis.exec.IExecChainContext) IDataxReader(com.qlangtech.tis.datax.IDataxReader) DataxProcessor(com.qlangtech.tis.datax.impl.DataxProcessor) RunningStatus(com.qlangtech.tis.fullbuild.indexbuild.RunningStatus) ExecuteResult(com.qlangtech.tis.exec.ExecuteResult) IRemoteTaskTrigger(com.qlangtech.tis.fullbuild.indexbuild.IRemoteTaskTrigger) IDataxWriter(com.qlangtech.tis.datax.IDataxWriter) JoinPhaseStatus(com.qlangtech.tis.fullbuild.phasestatus.impl.JoinPhaseStatus) TISReactor(com.qlangtech.tis.fullbuild.taskflow.TISReactor) ReactorListener(org.jvnet.hudson.reactor.ReactorListener) File(java.io.File)

Aggregations

Lists (com.google.common.collect.Lists)1 Maps (com.google.common.collect.Maps)1 FullbuildPhase (com.qlangtech.tis.assemble.FullbuildPhase)1 DataXJobSubmit (com.qlangtech.tis.datax.DataXJobSubmit)1 IDataXBatchPost (com.qlangtech.tis.datax.IDataXBatchPost)1 IDataxReader (com.qlangtech.tis.datax.IDataxReader)1 IDataxWriter (com.qlangtech.tis.datax.IDataxWriter)1 DataxProcessor (com.qlangtech.tis.datax.impl.DataxProcessor)1 ExecuteResult (com.qlangtech.tis.exec.ExecuteResult)1 IExecChainContext (com.qlangtech.tis.exec.IExecChainContext)1 TrackableExecuteInterceptor (com.qlangtech.tis.exec.impl.TrackableExecuteInterceptor)1 IRemoteTaskTrigger (com.qlangtech.tis.fullbuild.indexbuild.IRemoteTaskTrigger)1 RunningStatus (com.qlangtech.tis.fullbuild.indexbuild.RunningStatus)1 DumpPhaseStatus (com.qlangtech.tis.fullbuild.phasestatus.impl.DumpPhaseStatus)1 JoinPhaseStatus (com.qlangtech.tis.fullbuild.phasestatus.impl.JoinPhaseStatus)1 DataflowTask (com.qlangtech.tis.fullbuild.taskflow.DataflowTask)1 TISReactor (com.qlangtech.tis.fullbuild.taskflow.TISReactor)1 ISelectedTab (com.qlangtech.tis.plugin.ds.ISelectedTab)1 IncrStatusUmbilicalProtocol (com.qlangtech.tis.realtime.yarn.rpc.IncrStatusUmbilicalProtocol)1 AdapterStatusUmbilicalProtocol (com.qlangtech.tis.realtime.yarn.rpc.impl.AdapterStatusUmbilicalProtocol)1