elasticsearch-jdbc 1.7.3 执行流程（一）

• elasticsearch-jdbc启动方式

java \
    -cp "/opt/elasticsearch-jdbc/lib/*" \
    -Dlog4j.configurationFile=/opt/elasticsearch-jdbc/bin/log4j2.xml \
    org.xbib.tools.Runner \
    org.xbib.tools.JDBCImporter \
    /tmp/config.json

• 执行org.xbib.tools.Runner的main方法

public static void main(String[] args) {
        try {
            Class clazz = Class.forName(args[0]); //获取org.xbib.tools.JDBCImporter类
            CommandLineInterpreter commandLineInterpreter = (CommandLineInterpreter) clazz.newInstance();
            InputStream in = args.length > 1 ? new FileInputStream(args[1]) : System.in; //获取config.json配置文件流
            commandLineInterpreter.reader("args", in).run(true); 执行JDBCImporter的run(boolean bootstrap)方法
            in.close();
        } catch (Throwable e) {
            e.printStackTrace();
            System.exit(1);
        }
        System.exit(0);
    }

• JDBCImporter的run方法：

  prepare(); //初始化elasticsearch 客户端 创建索引
            futures = schedule(settings);//根据config来判断是否配置了定时任务。如何设置了，使用ScheduledThreadPoolExecutor定时执行
            if (!futures.isEmpty()) {
                logger.debug("waiting for {} futures...", futures.size());
                for (Future future : futures) {
                    try {
                        Object o = future.get();
                        logger.debug("got future {}", o);
                    } catch (CancellationException e) {
                        logger.warn("schedule canceled");
                    } catch (InterruptedException e) {
                        logger.error(e.getMessage(), e);
                    } catch (ExecutionException e) {
                        logger.error(e.getMessage(), e);
                    }
                }
                logger.debug("futures complete");
            } else {
                execute(); //否则就只执行一次
            }

• execute方法实例化MetricSimplePipelineExecutor，并执行execute方法。而execute实质上执行的是JDBCImporter的call方法。

 public MeterMetric call() throws Exception {
        try {
            metric = new MeterMetric(5L, TimeUnit.SECONDS);
            Iterator<R> it = this;
            while (it.hasNext()) {
                R r = it.next();
                // add ourselves if not already done
                requestListeners.put(null, this); //添加当前类 JDBCImporter，JDBCImporter实现了PipelineRequestListener
                errorListeners.put(null, this);
                for (PipelineRequestListener<MeterMetric, R> requestListener : requestListeners.values()) {
                    try {
                        requestListener.newRequest(this, r); //执行newRequest方法，实质上执行的是JDBCImporter的process(Settings settings)方法
                    } catch (PipelineException e) {
                        for (PipelineErrorListener<MeterMetric, R, E> errorListener : errorListeners.values()) {
                            errorListener.error(this, r, (E) e);
                        }
                    }
                }
                metric.mark();
            }
            close();
        } finally {
            metric.stop();
        }
        return getMetric();
    }

• JDBCImporter的process方法初始化StandardContext，并执行StandardContext的execute方法

public void execute() throws Exception {
        try {
            state = State.BEFORE_FETCH;
            beforeFetch();
            state = State.FETCH;
            fetch(); //抓取数据源的数据
        } finally {
            state = State.AFTER_FETCH;
            afterFetch();//将数据发送给es
            state = State.IDLE;
        }
    }