CliFronted是FLink提交任务的入口以及其他命令如取消等命令入口

    提交流程分析：

    1.    public CliFrontend(
    2.             Configuration configuration,
    3.             ClusterClientServiceLoader clusterClientServiceLoader,
    4.             List<CustomCommandLine> customCommandLines) {
    5.         this.configuration = checkNotNull(configuration);
    6.         this.customCommandLines = checkNotNull(customCommandLines);
    7.         this.clusterClientServiceLoader = checkNotNull(clusterClientServiceLoader);
    9.         FileSystem.initialize(
    10.                 configuration, PluginUtils.createPluginManagerFromRootFolder(configuration));
    12.         this.customCommandLineOptions = new Options();
    14.         for (CustomCommandLine customCommandLine : customCommandLines) {
    15.             customCommandLine.addGeneralOptions(customCommandLineOptions);
    16.             customCommandLine.addRunOptions(customCommandLineOptions);
    17.         }
    19.         this.clientTimeout = configuration.get(ClientOptions.CLIENT_TIMEOUT);
    20.         this.defaultParallelism = configuration.getInteger(CoreOptions.DEFAULT_PARALLELISM);
    21.     }

    ClusterClientServiceLoader**：SPI**加载对应的ClusterClientFactory，默认是 org.apache.flink.client.deployment.StandaloneClientFactory

    customCommandLines：默认含有 Generic Default两种，获取active的CommandLine，默认是Default
    Generic 是run-application模式，1.11开始出现https://blog.csdn.net/yuchuanchen/article/details/107617738
    不同的customCommandLines最终转换得到的Configuration是不同的

    PackagedProgram：相关的配置信息

    • jar包
    • classPath
    • 入口类名称
    • Configuration
    • mainClass
    • userCodeClassLoader
    • ……


    DefaultExecutorServiceLoader： SPI加载对应的**PipelineExecutorFactory，默认是org.apache.flink.client.deployment.executors.RemoteExecutorFactory
    org.apache.flink.client.deployment.executors.LocalExecutorFactory

    设置环境Context

    1.  ContextEnvironment.setAsContext(
    2.                     executorServiceLoader,
    3.                     configuration,
    4.                     userCodeClassLoader,
    5.                     enforceSingleJobExecution,
    6.                     suppressSysout);
    8.             StreamContextEnvironment.setAsContext(
    9.                     executorServiceLoader,
    10.                     configuration,
    11.                     userCodeClassLoader,
    12.                     enforceSingleJobExecution,
    13.                     suppressSysout);

    执行客户Main方法


    **