hbase跨机房同步

Dear 丶 2024-04-18 08:42 38阅读 0赞

两边hbase版本均是1.1.4，两个hbase集群之间的数据同步  
hbase通过命令看数据，太繁琐，可以[hbaseclient][]  
参考 [HBase备份还原OpenTSDB数据之Snapshot][HBase_OpenTSDB_Snapshot]和[Hbase四种数据迁移方案][Hbase]  
**1 创建快照**

hbase shell
    snapshot 'your_table_snapshot',’your_table'

**2 在另一个集群中快照中恢复数据**  
执行命令可以查看快照清单

[root@bwsc65 ~]# hadoop fs -ls hdfs://172.19.123.151:9000/hbase/.hbase-snapshot/
    SLF4J: Class path contains multiple SLF4J bindings.
    SLF4J: Found binding in [jar:file:/application/hadoop-2.6.4/share/hadoop/common/lib/slf4j-log4j12-1.7.5.jar!/org/slf4j/impl/StaticLoggerBinder.class]
    SLF4J: Found binding in [jar:file:/application/hbase-1.1.4/lib/slf4j-log4j12-1.7.5.jar!/org/slf4j/impl/StaticLoggerBinder.class]
    SLF4J: See http://www.slf4j.org/codes.html#multiple_bindings for an explanation.
    SLF4J: Actual binding is of type [org.slf4j.impl.Log4jLoggerFactory]
    Found 2 items
    drwxr-xr-x   - root supergroup          0 2019-09-04 14:07 hdfs://172.19.123.151:9000/hbase/.hbase-snapshot/.tmp
    drwxr-xr-x   - root supergroup          0 2019-09-04 14:07 hdfs://172.19.123.151:9000/hbase/.hbase-snapshot/my_table_09041407

在目的端hbase集群中，执行下面的命令进行快照的复制

hbase org.apache.hadoop.hbase.snapshot.ExportSnapshot \
    -snapshot my_table_09041407 -copy-from hdfs://172.19.123.151:9000/hbase \
    -copy-to hdfs://10.101.10.65:9000/hbase   -overwrite -mappers 16 -bandwidth  1024

提示异常,java堆内存不够

2019-09-04 14:19:00,993 INFO  [main] mapreduce.Job: Task Id : attempt_1558491925397_0127_m_000013_2, Status : FAILED
    Error: Java heap space
    2019-09-04 14:19:01,995 INFO  [main] mapreduce.Job:  map 100% reduce 0%
    2019-09-04 14:19:02,001 INFO  [main] mapreduce.Job: Job job_1558491925397_0127 failed with state FAILED due to: Task failed task_1558491925397_0127_m_000002
    Job failed as tasks failed. failedMaps:1 failedReduces:0
    
    2019-09-04 14:19:02,070 INFO  [main] mapreduce.Job: Counters: 12
            Job Counters
                    Failed map tasks=41
                    Killed map tasks=15
                    Launched map tasks=51
                    Other local map tasks=51
                    Total time spent by all maps in occupied slots (ms)=474640
                    Total time spent by all reduces in occupied slots (ms)=0
                    Total time spent by all map tasks (ms)=118660
                    Total vcore-seconds taken by all map tasks=118660
                    Total megabyte-seconds taken by all map tasks=121507840
            Map-Reduce Framework
                    CPU time spent (ms)=0
                    Physical memory (bytes) snapshot=0
                    Virtual memory (bytes) snapshot=0
    2019-09-04 14:19:02,072 ERROR [main] snapshot.ExportSnapshot: Snapshot export failed
    org.apache.hadoop.hbase.snapshot.ExportSnapshotException: Copy Files Map-Reduce Job failed
            at org.apache.hadoop.hbase.snapshot.ExportSnapshot.runCopyJob(ExportSnapshot.java:804)
            at org.apache.hadoop.hbase.snapshot.ExportSnapshot.run(ExportSnapshot.java:997)
            at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:70)
            at org.apache.hadoop.hbase.snapshot.ExportSnapshot.innerMain(ExportSnapshot.java:1071)
            at org.apache.hadoop.hbase.snapshot.ExportSnapshot.main(ExportSnapshot.java:1075)

有人说可以通过修改hadoop的配置

vim mapred-site.xml
    # 添加以下内容
            
            <property>
                    <name>mapred.child.java.opts</name>
                    <value>-Xmx1024m</value>
             </property>
     stop-hbase.sh 
     stop-all.sh
     start-all.sh
     start-hbase.sh

可是我的服务器在云端中，内存已经不足了，增加要钱，该怎么办呢？  
我换了中策略,有点慢，但不会立马提示异常信息。

hadoop distcp -Dmapreduce.job.queue.name=queue_0001_01 -update -skipcrccheck -m 100 hdfs://172.19.123.151:9000/hbase/data/default/my_table /hbase/data/default/my_table

我等了一个晚上，卡在这里不动了。这是为什么呢？  
![1][]  
**3 copyTable**  
`copyTable`需要在1T以下还可以，1T以上的表就不能用这种方式，在执行前，需要将表构建好，同时考虑hbase的版本，我这里版本是一致的

# 这个在目标服务器上执行，
    create '表', { NAME => 'o', COMPRESSION => 'SNAPPY' }, {NUMREGIONS => 10, SPLITALGO => 'HexStringSplit'}
    
    # 这个在源服务器上执行
    # *.test.com.cn这些是目标服务器上zookeeper的，这里不是走的vpn，因为没考虑这个预算，故只能设置白名单的方式控制访问。
    hbase org.apache.hadoop.hbase.mapreduce.CopyTable --peer.adr=bg15.test.com.cn,bg7.test.com.cn,bg16.test.com.cn,bg6.test.com.cn,bg8.test.com.cn:2181:/hbase 表

2351万数据耗费了1h11m，云上带宽是1MB/s  
![2][]  
在有限的资源下把事情搞成，而没有采用华为服务人员说的CDM，找的代理商都没有使用过，就推销给客户使用，那我岂不是成了小白鼠。有些事情还是自己经历，才算完整。  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dhcnJhaA_size_16_color_FFFFFF_t_70]

[hbaseclient]: https://github.com/HY-ZhengWei/HBaseClient
[HBase_OpenTSDB_Snapshot]: https://www.cnblogs.com/yybrhr/p/11165652.html
[Hbase]: https://blog.csdn.net/TXBSW/article/details/88173304
[1]: https://image.dandelioncloud.cn/pgy_files/images/2024/04/17/e3c0db1d3a334a8498b4d7e6035676ef.png
[2]: https://image.dandelioncloud.cn/pgy_files/images/2024/04/17/2d99c3dfea3d4a8b87a8c54638d2f96f.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dhcnJhaA_size_16_color_FFFFFF_t_70]: https://image.dandelioncloud.cn/pgy_files/images/2024/04/17/c655fb172e9a4dadb3f998d4e8442b9d.png