kafka connect，将数据批量写到hdfs完整过程

bin/connect-standalone.sh config/connect-standalone.properties config/connector1.properties

name="test"    #该connector的名字

#将自己按connect接口规范编写的代码打包后放在kafka/libs目录下，再根据项目结构引用对应connector
connector.class=hdfs.HdfsSinkConnector 

#Task是导入导出的具体实现，这里是指定多少个task来并行运行导入导出作业，由多线程实现。由于hdfs中一个文件每次只能又一个文件操作，所以这里只能是1

tasks.max=1 

#指定从哪个topic读取数据，这些其实是用来在connector或者task的代码中读取的。

topics=test 

#指定key以那种方式转换，需和Producer发送方指定的序列化方式一致

key.converter=org.apache.kafka.connect.converters.ByteArrayConverter 
value.converter=org.apache.kafka.connect.json.JsonConverter     #同上

hdfs.url=hdfs://127.0.0.1:9000　　#hdfs的url路径，在Connector中会被读取
hdfs.path=/test/file　　#hdfs文件路径，同样Connector中被读取

key.converter.schemas.enable=true　　#稍后介绍，可以true也可以false，影响传输格式
value.converter.schemas.enable=true　　#稍后介绍，可以true也可以false

public class HdfsSinkConnector extends SinkConnector {
    //这两项为配置hdfs的urlh和路径的配置项，需要在connector1.properties中指定
    public static final String HDFS_URL = "hdfs.url";
    public static final String HDFS_PATH = "hdfs.path";
    private static final ConfigDef CONFIG_DEF = new ConfigDef()
            .define(HDFS_URL, ConfigDef.Type.STRING, ConfigDef.Importance.HIGH, "hdfs url")
            .define(HDFS_PATH, ConfigDef.Type.STRING, ConfigDef.Importance.HIGH, "hdfs path");
    private String hdfsUrl;
    private String hdfsPath;
    @Override
    public String version() {
        return AppInfoParser.getVersion();
    }

//start方法会再初始的时候执行一次，这里主要用于配置
    @Override
    public void start(Map<String, String> props) {
        hdfsUrl = props.get(HDFS_URL);
        hdfsPath = props.get(HDFS_PATH);
    }
　　//这里指定了Task的类
    @Override
    public Class<? extends Task> taskClass() {
        return HdfsSinkTask.class;
    }
　　//用于配置Task的config，这些都是会在Task中用到
    @Override
    public List<Map<String, String>> taskConfigs(int maxTasks) {
        ArrayList<Map<String, String>> configs = new ArrayList<>();
        for (int i = 0; i < maxTasks; i++) {
            Map<String, String> config = new HashMap<>();
            if (hdfsUrl != null)
                config.put(HDFS_URL, hdfsUrl);
            if (hdfsPath != null)
                config.put(HDFS_PATH, hdfsPath);
            configs.add(config);
        }
        return configs;
    }
　　//关闭时的操作，一般是关闭资源。
    @Override
    public void stop() {
        // Nothing to do since FileStreamSinkConnector has no background monitoring.
    }

    @Override
    public ConfigDef config() {
        return CONFIG_DEF;
    }

}

public class HdfsSinkTask extends SinkTask {
    private static final Logger log = LoggerFactory.getLogger(HdfsSinkTask.class);

    private String filename;

    public static String hdfsUrl;
    public static String hdfsPath;
    private Configuration conf;
    private FSDataOutputStream os;
    private FileSystem hdfs;


    public HdfsSinkTask(){

    }

    @Override
    public String version() {
        return new HdfsSinkConnector().version();
    }
　　//Task开始会执行的代码，可能有多个Task，所以每个Task都会执行一次
    @Override
    public void start(Map<String, String> props) {
        hdfsUrl = props.get(HdfsSinkConnector.HDFS_URL);
        hdfsPath = props.get(HdfsSinkConnector.HDFS_PATH);
        System.out.println("----------------------------------- start--------------------------------");

        conf = new Configuration();
        conf.set("fs.defaultFS", hdfsUrl);
        //这两个是与hdfs append相关的设置
        conf.setBoolean("dfs.support.append", true);
        conf.set("dfs.client.block.write.replace-datanode-on-failure.policy", "NEVER");
        try{
            hdfs = FileSystem.get(conf);
//            connector.hdfs = new Path(HDFSPATH).getFileSystem(conf);
            os = hdfs.append(new Path(hdfsPath));
        }catch (IOException e){
            System.out.println(e.toString());
        }

    }
　　//核心操作，put就是将数据从kafka中取出，存放到其他地方去
    @Override
    public void put(Collection<SinkRecord> sinkRecords) {
        for (SinkRecord record : sinkRecords) {
            log.trace("Writing line to {}: {}", logFilename(), record.value());
            try{
                System.out.println("write info------------------------" + record.value().toString() + "-----------------");
                os.write((record.value().toString()).getBytes("UTF-8"));
                os.hsync();
            }catch(Exception e){
                System.out.print(e.toString());
            }
        }
    }

    @Override
    public void flush(Map<TopicPartition, OffsetAndMetadata> offsets) {
        try{
            os.hsync();
        }catch (Exception e){
            System.out.print(e.toString());
        }

    }

//同样是结束时候所执行的代码，这里用于关闭hdfs资源
    @Override
    public void stop() {
        try {
            os.close();
        }catch(IOException e){
            System.out.println(e.toString());
        }
    }

    private String logFilename() {
        return filename == null ? "stdout" : filename;
    }


}

{
    "schema": {
        "type": "struct",
        "fields": [{
            "type": "int32",
            "optional": true,
            "field": "c1"
        }, {
            "type": "string",
            "optional": true,
            "field": "c2"
        }, {
            "type": "int64",
            "optional": false,
            "name": "org.apache.kafka.connect.data.Timestamp",
            "version": 1,
            "field": "create_ts"
        }, {
            "type": "int64",
            "optional": false,
            "name": "org.apache.kafka.connect.data.Timestamp",
            "version": 1,
            "field": "update_ts"
        }],
        "optional": false,
        "name": "foobar"
    },
    "payload": {
        "c1": 10000,
        "c2": "bar",
        "create_ts": 1501834166000,
        "update_ts": 1501834166000
    }
}  

org.apache.kafka.connect.errors.DataException: JsonConverter with schemas.enable requires "schema" and "payload" fields and may not contain additional fields. If you are trying to deserialize plain JSON data, set schemas.enable=false in your converter configuration.

   at org.apache.kafka.connect.json.JsonConverter.toConnectData(JsonConverter.java:308)