线上商城运营方案,平板电视seo优化关键词,护肤品网页设计图片,用超轻粘土做网站文章目录 一.创建RecordWriter实例都做了啥1. 根据recordWrites数量创建不同的代理类2. 创建RecordWriters3. 单个RecordWriter的创建细节 二. RecordWriter包含的主要组件1. RecordWriter两种实现类分别实现分发策略和广播2. ChannelSelectorRecordWriter的发送策略2.1. Chann… 文章目录 一.创建RecordWriter实例都做了啥1. 根据recordWrites数量创建不同的代理类2. 创建RecordWriters3. 单个RecordWriter的创建细节 二. RecordWriter包含的主要组件1. RecordWriter两种实现类分别实现分发策略和广播2. ChannelSelectorRecordWriter的发送策略2.1. ChannelSelector根据实现类实现不同发送策略 2. BroadcastRecordWriter广播所有元素到下游InputChannel StreamTask节点中的中间结果数据元素最终通过RecordWriterOutput实现了网络输出RecordWriterOutput底层依赖RecordWriter组件完成数据输出操作接下来我们深入了解RecordWriter的设计和实现。
一.创建RecordWriter实例都做了啥
在StreamTask构造器方法中会直接创建RecordWriter实例用于输出当前任务产生的Intermediate Result数据。
1. 根据recordWrites数量创建不同的代理类
其中createRecordWriterDelegate根据recordWrites的数量创建对应的RecordWriterDelegate代理类。
protected StreamTask( Environment environment, Nullable TimerService timerService, Thread.UncaughtExceptionHandler uncaughtExceptionHandler, StreamTaskActionExecutor actionExecutor, TaskMailbox mailbox) throws Exception {...this.recordWriter createRecordWriterDelegate(configuration, environment);...}
//1. 如果recordWrites数量等于1则创建SingleRecordWriter代理类
//2. 如果recordWrites数量等于0则创建NonRecordWriter代理类
//3. 其他情况则创建MultipleRecordWriters代理类。VisibleForTesting
public static OUT RecordWriterDelegateSerializationDelegateStreamRecordOUT createRecordWriterDelegate( StreamConfig configuration, Environment environment) { ListRecordWriterSerializationDelegateStreamRecordOUT recordWrites createRecordWriters(configuration, environment); if (recordWrites.size() 1) { return new SingleRecordWriter(recordWrites.get(0)); } else if (recordWrites.size() 0) { return new NonRecordWriter(); } else { return new MultipleRecordWriters(recordWrites); }
}2. 创建RecordWriters
获取StreamTask的所有输出边放到RecordWriters中返回创建的RecordWriter集合。
private static OUT ListRecordWriterSerializationDelegateStreamRecordOUT createRecordWriters(StreamConfig configuration,Environment environment) {// 创建RecordWriter集合ListRecordWriterSerializationDelegateStreamRecordOUT recordWriters new ArrayList();// 获取输出的StreamEdge(所有的输出边)ListStreamEdge outEdgesInOrder configuration.getOutEdgesInOrder(environment.getUserClassLoader());// 获取chainedConfigs参数MapInteger, StreamConfig chainedConfigs configuration.getTransitiveChainedTaskConfigsWithSelf(environment.getUserClassLoader());// 遍历输出节点分别创建RecordWriter实例for (int i 0; i outEdgesInOrder.size(); i) {StreamEdge edge outEdgesInOrder.get(i);recordWriters.add(createRecordWriter(edge,i,environment,environment.getTaskInfo().getTaskName(),chainedConfigs.get(edge.getSourceId()).getBufferTimeout()));}return recordWriters;
}3. 单个RecordWriter的创建细节
看单个RecordWriter的创建过程包括分区策略、缓存结果的ResultPartition、最后通过RecordWriterBuilder创建RecordWriter、以及设定MetricGroup监控RecordWriter指标并输出。 创建outputPartitionerStreamPartitioner分区策略会被应用在RecordWriter中例如DataStream.rebalance()操作就会创建RebalancePartitioner作为StreamPartitioner的实现类并通过RebalancePartitioner选择下游InputChannel实现数据元素按照指定的分区策略下发。ResultPartition ResultPartition内部会在本地存储需要下发的Buffer数据并等待下游节点向上游节点发送数据消费请求。通过RecordWriterBuilder创建RecordWriter在创建过程中会设定outputPartitioner、bufferTimeout以及bufferWriter等参数。为RecordWriter设定MetricGroup用于监控指标的采集和输出。 private static OUT RecordWriterSerializationDelegateStreamRecordOUT createRecordWriter(StreamEdge edge,int outputIndex,Environment environment,String taskName,long bufferTimeout) {SuppressWarnings(unchecked)// 获取边上的StreamPartitionerStreamPartitionerOUT outputPartitioner (StreamPartitionerOUT) edge.getPartitioner();LOG.debug(Using partitioner {} for output {} of task {}, outputPartitioner, outputIndex, taskName);// 获取ResultPartitionWriterResultPartitionWriter bufferWriter environment.getWriter(outputIndex);// 初始化Partitionerif (outputPartitioner instanceof ConfigurableStreamPartitioner) {int numKeyGroups bufferWriter.getNumTargetKeyGroups();if (0 numKeyGroups) {((ConfigurableStreamPartitioner) outputPartitioner).configure(numKeyGroups);}}// 创建RecordWriterRecordWriterSerializationDelegateStreamRecordOUT output new RecordWriterBuilderSerializationDelegateStreamRecordOUT().setChannelSelector(outputPartitioner).setTimeout(bufferTimeout).setTaskName(taskName).build(bufferWriter);// 设定MetricGroup监控output.setMetricGroup(environment.getMetricGroup().getIOMetricGroup());return output;
}二. RecordWriter包含的主要组件
RecordWriter内部主要包含RecordSerializer和ResultPartitionWriter两个组件。 RecordSerializer用于对输出到网络中的数据进行序列化操作将数据元素序列化成Bytes[]二进制格式维护Bytes[]数据中的startBuffer及position等信息。ResultPartitionWriter是ResultPartition实现的接口提供了将数据元素写入ResultPartiton的方法例如addBufferConsumer()方法就是将RecordSerializer序列化的BufferConsumer数据对象添加到ResultPartition队列并进行缓存供下游InputGate消费BufferConsumer对象。 1. RecordWriter两种实现类分别实现分发策略和广播
RecordWriter主要有两种实现类ChannelSelectorRecordWriter和BroadcastRecordWriter。 ChannelSelectorRecordWriter根据ChannelSelector选择下游节点的InputChannelChannelSelector内部基于StreamPartitoner获取不同的数据下发策略最终实现数据重分区。BroadcastRecordWriter对应广播式数据下发即数据元素会被发送到下游所有的InputChannel中。当用户执行了Broadcast操作时就会创建BroadcastRecordWriter实现数据元素的广播下发操作。 如代码通过RecordWriterBuilder创建RecordWriter此时会根据selector.isBroadcast()条件选择创建ChannelSelectorRecordWriter还是BroadcastRecordWriter实例。
public RecordWriterT build(ResultPartitionWriter writer) {if (selector.isBroadcast()) {return new BroadcastRecordWriter(writer, timeout, taskName);} else {return new ChannelSelectorRecordWriter(writer, selector, timeout, taskName);}
} 2. ChannelSelectorRecordWriter的发送策略
ChannelSelectorRecordWriter控制数据元素发送到下游的哪些InputChannel中。如代码调用channelSelector.selectChannel(record)选择下游的InputChannel。
对于非广播类型的分区器最终都会创建ChannelSelectorRecordWriter实现StreamRecord数据的下发操作。
public void emit(T record) throws IOException, InterruptedException {emit(record, channelSelector.selectChannel(record));
}2.1. ChannelSelector根据实现类实现不同发送策略
ChannelSelector的实现类主要有StreamPartitioner、RoundRobinChannelSelector和OutputEmitter三种。 StreamPartitionerDataStream API中物理操作指定的分区器例如当用户调用DataStream.rebalance()方法时会创建RebalencePartitioner。在StreamTask执行的过程中会获取相应的StreamPartitioner应用在ChannelSelectorRecordWriter中实现对数据元素分区的选择。RoundRobinChannelSelectorChannelSelector的默认实现类提供了对Round-Robin策略的支持以轮询的方式随机选择一个分区输出数据元素。OutputEmitter适用于BatchTask须配合ShipStrategyType使用通过ShipStrategyType执行的策略输出数据。 2. BroadcastRecordWriter广播所有元素到下游InputChannel
BroadcastRecordWriter的实现就比较简单了在BroadcastRecordWriter中不需要ChannelSelector组件选择数据元素分区直接将所有的数据元素广播发送到下游所有InputChannel中即可。