ITPub博客
首页 博客 专栏 学院 社区 IT新媒体评选
发博

首页 > 大数据 > Hadoop > Hadoop2.7实战v1.0之HDFS HA

Hadoop2.7实战v1.0之HDFS HA

原创 Hadoop 作者：hackeruncle 时间：2016-03-06 18:07:30 0 删除 编辑

HDFS HA实战v1.0


当前环境:hadoop+zookeeper(namenode,resourcemanager HA)
 namenode  serviceId  Init status
 sht-sgmhadoopnn-01  nn1  active
 sht-sgmhadoopnn-02  nn2  standby

参考: http://blog.csdn.net/u011414200/article/details/50336735

一.查看namenode是active还是standby

1.打开网页


2.查看zkfc日志 

点击(此处)折叠或打开

  1. [root@sht-sgmhadoopnn-01 logs]# more hadoop-root-zkfc-sht-sgmhadoopnn-01.telenav.cn.log

  3. …………………..

  5. 2016-02-28 00:24:00,692 INFO org.apache.hadoop.ha.ZKFailoverController: Trying to make NameNode at sht-sgmhadoopnn-01/172.16.101.55:8020 active...

  7. 2016-02-28 00:24:01,762 INFO org.apache.hadoop.ha.ZKFailoverController: Successfully transitioned NameNode at sht-sgmhadoopnn-01/172.16.101.55:8020 to active state
  8.  

  10. [root@sht-sgmhadoopnn-02 logs]# more hadoop-root-zkfc-sht-sgmhadoopnn-01.telenav.cn.log

  12. …………………..

  14. 2016-02-28 00:24:01,186 INFO org.apache.hadoop.ha.ZKFailoverController: ZK Election indicated that NameNode at sht-sgmhadoopnn-02/172.16.101.56:8020 should become standby

  16. 2016-02-28 00:24:01,209 INFO org.apache.hadoop.ha.ZKFailoverController: Successfully transitioned NameNode at sht-sgmhadoopnn-02/172.16.101.56:8020 to standby state


 3. 通过命令hdfs haadmin –getServiceState    

###$HADOOP_HOME/etc/hadoop/hdfs-site.xml, dfs.ha.namenodes.[dfs.nameservices]

               

                                <!--设置NameNode IDs 此版本最大只支持两个NameNode -->

                                dfs.ha.namenodes.mycluster

                                nn1,nn2

               


 

点击(此处)折叠或打开

  1. [root@sht-sgmhadoopnn-02 logs]# hdfs haadmin -getServiceState nn1

  3. active

  5.  [root@sht-sgmhadoopnn-02 logs]# hdfs haadmin -getServiceState nn2

  7. standby
 

二.基本命令

点击(此处)折叠或打开

  1. [root@sht-sgmhadoopnn-02 logs]# hdfs --help

  3. Usage: hdfs [--config confdir] [--loglevel loglevel] COMMAND

  5.        where COMMAND is one of:

  7.   dfs run a filesystem command on the file systems supported in Hadoop.

  9.   classpath prints the classpath

  11.   namenode -format format the DFS filesystem

  13.   secondarynamenode run the DFS secondary namenode

  15.   namenode run the DFS namenode

  17.   journalnode run the DFS journalnode

  19.   zkfc run the ZK Failover Controller daemon

  21.   datanode run a DFS datanode

  23.   dfsadmin run a DFS admin client

  25.   haadmin run a DFS HA admin client

  27.   fsck run a DFS filesystem checking utility

  29.   balancer run a cluster balancing utility

  31.   jmxget get JMX exported values from NameNode or DataNode.

  33.   mover run a utility to move block replicas across

  35.                                storage types

  37.   oiv apply the offline fsimage viewer to an fsimage

  39.   oiv_legacy apply the offline fsimage viewer to an legacy fsimage

  41.   oev apply the offline edits viewer to an edits file

  43.   fetchdt fetch a delegation token from the NameNode

  45.   getconf get config values from configuration

  47.   groups get the groups which users belong to

  49.   snapshotDiff diff two snapshots of a directory or diff the

  51.                        current directory contents with a snapshot

  53.   lsSnapshottableDir list all snapshottable dirs owned by the current user

  55.                                                 Use -help to see options

  57.   portmap run a portmap service

  59.   nfs3 run an NFS version 3 gateway

  61.   cacheadmin configure the HDFS cache

  63.   crypto configure HDFS encryption zones

  65.   storagepolicies list/get/set block storage policies

  67.   version print the version

  69. ###########################################################################

  71. [root@sht-sgmhadoopnn-02 logs]# hdfs namenode --help

  73. Usage: java NameNode [-backup] |

  75.         [-checkpoint] |

  77.         [-format [-clusterid cid ] [-force] [-nonInteractive] ] |

  79.         [-upgrade [-clusterid cid] [-renameReserved<k-v pairs>] ] |

  81.         [-upgradeOnly [-clusterid cid] [-renameReserved<k-v pairs>] ] |

  83.         [-rollback] |

  85.         [-rollingUpgrade <rollback|downgrade|started> ] |

  87.         [-finalize] |

  89.         [-importCheckpoint] |

  91.         [-initializeSharedEdits] |

  93.         [-bootstrapStandby] |

  95.         [-recover [ -force] ] |

  97.         [-metadataVersion ] ]




  102. ###########################################################################

  104. [root@sht-sgmhadoopnn-02 logs]# hdfs haadmin --help

  106. -help: Unknown command

  108. Usage: haadmin

  110.     [-transitionToActive [--forceactive] <serviceId>]

  112.     [-transitionToStandby <serviceId>]

  114.     [-failover [--forcefence] [--forceactive] <serviceId> <serviceId>]

  116.     [-getServiceState <serviceId>]

  118.     [-checkHealth <serviceId>]

  120.     [-help <command>]
transitionToActive和transitionToStandby是用于在不同状态之间切换的。 

failover 初始化一个故障恢复。该命令会从一个失效的NameNode切换到另一个上面(不支持在自动切换的环境操作)。

getServiceState 获取当前NameNode的状态。

checkHealth 检查NameNode的状态。正常就返回0，否则返回非0值。  

 

三.实验

1.测试HDFS的手工切换功能(失败)

点击(此处)折叠或打开

  1. [root@sht-sgmhadoopnn-01 ~]# hdfs haadmin -failover --forceactive nn1 nn2

  3. forcefence and forceactive flags not supported with auto-failover enabled.

#hdfs-site.xml 中设置dfs.ha.automatic-failover.enabled为 true,故提示不能手动切换

 

2.测试HDFS的自动切换功能(成功)

   a.在active namenode机器上通过jps命令查找到namenode的进程号，然后通过kill -9的方式杀掉进程，观察另一个namenode节点是否会从状态standby变成active状态

点击(此处)折叠或打开

  1. [root@sht-sgmhadoopnn-01 ~]# hdfs haadmin -getServiceState nn1

  3. active

  5. [root@sht-sgmhadoopnn-01 ~]# hdfs haadmin -getServiceState nn2

  7. standby




  12. [root@sht-sgmhadoopnn-01 ~]# jps

  14. 10327 ResourceManager

  16. 10162 DFSZKFailoverController

  18. 9821 NameNode

  20. 20064 Jps

  22. [root@sht-sgmhadoopnn-01 ~]# kill -9 9821

  24. [root@sht-sgmhadoopnn-01 ~]# jps

  26. 10327 ResourceManager

  28. 10162 DFSZKFailoverController

  30. 20121 Jps

  32.  

  34. [root@sht-sgmhadoopnn-01 ~]# hdfs haadmin -getServiceState nn1

  36. 16/02/28 22:01:37 INFO ipc.Client: Retrying connect to server: sht-sgmhadoopnn-01/172.16.101.55:8020. Already tried 0 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=1, sleepTime=1000 MILLISECONDS)

  38. Operation failed: Call From sht-sgmhadoopnn-01.telenav.cn/172.16.101.55 to sht-sgmhadoopnn-01:8020 failed on connection exception: java.net.ConnectException: Connection refused; For more details see: http://wiki.apache.org/hadoop/ConnectionRefused

  40. [root@sht-sgmhadoopnn-01 ~]# hdfs haadmin -getServiceState nn2

  42. active

  44. [root@sht-sgmhadoopnn-01 ~]#

##也可以通过观察zkfc日志或者http://172.16.101.56:50070/ 来查看是否切换成功
 

   b.手动启动sht-sgmhadoopnn-01的namenode 

点击(此处)折叠或打开

  1. [root@sht-sgmhadoopnn-01 ~]# cd /hadoop/hadoop-2.7.2/sbin

  3. [root@sht-sgmhadoopnn-01 sbin]# jps

  5. 10327 ResourceManager

  7. 10162 DFSZKFailoverController

  9. 21640 Jps

  11. [root@sht-sgmhadoopnn-01 sbin]# hadoop-daemon.sh start namenode

  13. starting namenode, logging to /hadoop/hadoop-2.7.2/logs/hadoop-root-namenode-sht-sgmhadoopnn-01.telenav.cn.out

  15.  [root@sht-sgmhadoopnn-01 sbin]# jps

  17. 10327 ResourceManager

  19. 10162 DFSZKFailoverController

  21. 21784 Jps

  23. 21679 NameNode

  25. [root@sht-sgmhadoopnn-01 sbin]# hdfs haadmin -getServiceState nn1

  27. standby

  29. [root@sht-sgmhadoopnn-01 sbin]#

#### sht-sgmhadoopnn-01 机器上namenode 进程已经起来,且状态为standby

   c. 再次切换 

点击(此处)折叠或打开

  1. [root@sht-sgmhadoopnn-01 sbin]# hdfs haadmin -transitionToStandby nn2

  3. Automatic failover is enabled for NameNode at sht-sgmhadoopnn-02/172.16.101.56:8020

  5. Refusing to manually manage HA state, since it may cause

  7. a split-brain scenario or other incorrect state.

  9. If you are very sure you know what you are doing, please

  11. specify the --forcemanual flag.


  14. [root@sht-sgmhadoopnn-01 sbin]# hdfs haadmin -transitionToStandby --forcemanual nn2

  16. You have specified the --forcemanual flag. This flag is dangerous, as it can induce a split-brain scenario that WILL CORRUPT your HDFS namespace, possibly irrecoverably.

  18. It is recommended not to use this flag, but instead to shut down the cluster and disable automatic failover if you prefer to manually manage your HA state.

  20. You may abort safely by answering 'n' or hitting ^C now.

  22. Are you sure you want to continue? (Y or N) Y

  24. 16/02/28 22:30:22 WARN ha.HAAdmin: Proceeding with manual HA state management even though

  26. automatic failover is enabled for NameNode at sht-sgmhadoopnn-02/172.16.101.56:8020

  28.  

  30. [root@sht-sgmhadoopnn-01 sbin]# hdfs haadmin -getServiceState nn1

  32. active

  34. [root@sht-sgmhadoopnn-01 sbin]# hdfs haadmin -getServiceState nn2

  36. standby

  38. [root@sht-sgmhadoopnn-01 sbin]#


来自 “ ITPUB博客 ” ，链接：https://blog.itpub.net/30089851/viewspace-2047907/ ，如需转载，请注明出处，否则将追究法律责任。

0
0
上一篇： Hadoop2.7实战v1.0之添加DataNode节点后,更改文件复制策略dfs.replication
下一篇： Hadoop2.7实战v1.0之YARN HA
请登录后发表评论 登录
全部评论
hackeruncle
专家姓名：武基鹏
主要从事大数据平台产品的技术工作； 负责基于Spark、Flink生产开发及性能调优； 负责对公司的集群的部署、监控、调优和运营； 主导我司的数据平台0到1及实时中间件项目开发； 此外,精通Java、Shell、Python编程； 且荣获2016年IBM的Spark大赛全球第四名，中国区域第一名。

注册时间：2015-01-03

  • 博文量
    187
  • 访问量
    395984

最新文章

支持我们 作者招募 用户协议 FAQ Contact Us 编辑准则

北京盛拓优讯信息技术有限公司. 版权所有  京ICP备16024965号-8 北京市公安局海淀分局网监中心备案编号：11010802021510 niuxiaotong@pcpop.com 17352615567 未成年人举报专区

广播电视节目制作经营许可证(京) 字第1234号 中国互联网协会会员

违法和不良信息举报电话:18101376593