HDFS数据更新到hbase表

最新推荐文章于 2022-11-22 17:23:43 发布

weixin_33769207

最新推荐文章于 2022-11-22 17:23:43 发布

阅读量148

点赞数

CC 4.0 BY-SA版权

文章标签：大数据 python

原文链接：https://my.oschina.net/momisabuilder/blog/680923

本文详细介绍了如何使用自定义Mapper将HDFS文件高效地迁移至HBase中，并通过配置参数实现数据的批量导入与优化处理，确保数据一致性与性能优化。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

2019独角兽企业重金招聘Python工程师标准>>>

1，map类

public static class MyMapper extends Mapper<LongWritable, Text, NullWritable, Text> {

    List<Put> putList = new LinkedList<Put>();

    HTableInterface bvuser;

    public MyMapper() {

    }

  //这个方法每个map任务只调用一次
    @Override
    protected void setup(final Context context) throws IOException, InterruptedException {

        bvuser = new HTable(context.getConfiguration(),"bvuser");

        BufferedReader br = new BufferedReader(new InputStreamReader(FileSystem.get(context.getConfiguration()).open(new Path("hdfs文件路径"))));

        String line = null;

        while((line = br.readLine()) != null){
            
        }
        IOUtils.closeQuietly(br);

    }

  //hdfs的每行数据会调用一次
    @Override
    public void map(LongWritable key, Text value, Context context)throws IOException, InterruptedException {
        Counter counter = context.getCounter("group", "mapcount");
        counter.increment(1);

        String line = value.toString();
        if(line==null || line.isEmpty()){
            return;
        }

        Put put = new Put(Bytes.toBytes("3136949-" + userId));

        put.add(Bytes.toBytes("info"), Bytes.toBytes("billTypes"), System.currentTimeMillis(), Bytes.toBytes(StringUtils.join(typeidSet, ",")));

        putList.add(put);

        if(putList.size()==1000){
            bvuser.put(putList);
            putList.clear();
        }

    }

    //每个任务只掉用一次
    @Override
    protected void cleanup(Context context) throws IOException, InterruptedException {
        if(!putList.isEmpty()){
            bvuser.put(putList);
            putList.clear();
        }
        
        bvuser.close();
    }
}

2，Main方法

Configuration conf = HBaseConfiguration.create();
final Job job = Job.getInstance(conf, "data-save-job");

//格式化文件 文本形式 相当于字符串
job.setInputFormatClass(TextInputFormat.class);
job.setJarByClass(TextFile2HbaseJob.class);

//设置map类
job.setMapperClass(MyMapper.class);

//设置map的输出结果
job.setMapOutputKeyClass(NullWritable.class);
job.setMapOutputValueClass(Text.class);

FileOutputFormat.setOutputPath(job, new Path(output));
FileSystem fs = FileSystem.get(conf);
FileStatus[] files = fs.listStatus(new Path(input), new PathFilter() {
    @Override
    public boolean accept(Path path) {
        return path.getName().startsWith("part-m");
    }
});

if (files.length == 0) {
    throw new IllegalStateException("no file");
}
//整个目录下的文件都加进去 /export/part-m-0000000,part-m-0000000格式
for (FileStatus file : files) {
    FileInputFormat.addInputPath(job, file.getPath());
}
//设置reduce数0
job.setNumReduceTasks(0);

转载于:https://my.oschina.net/momisabuilder/blog/680923