四十一、SecondaryNameNode目录结构、nameNode故障恢复

深碍√TFBOYSˉ_ 2022-05-11 12:50 209阅读 0赞

#                         SecondaryNameNode目录结构、nameNode故障恢复 #

### 1、 SecondaryNameNode目录结构 ###

Secondary NameNode用来监控HDFS状态的辅助后台程序，每隔一段时间获取HDFS元数据的快照。  
在/opt/module/hadoop-2.7.2/data/tmp/dfs/namesecondary/current这个目录中查看SecondaryNameNode目录结构：

edits_0000000000000000001-0000000000000000002
    fsimage_0000000000000000002
    fsimage_0000000000000000002.md5
    VERSION

SecondaryNameNode的namesecondary/current目录和主namenode的current目录的布局相同。  
好处：在主namenode发生故障时（假设没有及时备份数据），可以从SecondaryNameNode恢复数据。

### 2、nameNode故障处理一：将SecondaryNameNode中数据拷贝到namenode存储数据的目录 ###

（1）kill -9 namenode进程号

[admin@hadoop14 current]$ jps
    5268 DataNode
    6409 Jps
    5545 NodeManager
    5131 NameNode
    [admin@hadoop14 current]$ kill -9 5131
    [admin@hadoop14 current]$ jps
    5268 DataNode
    5545 NodeManager
    6442 Jps

（2）删除namenode存储的数据（/opt/module/hadoop-2.7.2/data/tmp/dfs/name）

[admin@hadoop14 name]$ ll
    总用量 8
    drwxrwxr-x. 2 admin admin 4096 10月 20 16:18 current
    -rw-rw-r--. 1 admin admin   13 10月 20 14:49 in_use.lock
    [admin@hadoop14 name]$ rm -rf /opt/module/hadoop-2.7.2/data/tmp/dfs/name/*
    [admin@hadoop14 name]$ ll
    总用量 0

（3）拷贝SecondaryNameNode中数据到原namenode存储数据目录

[admin@hadoop16 namesecondary]$ rsync -rvl /opt/module/hadoop-2.7.2/data/tmp/dfs/namesecondary/* admin@hadoop14:/opt/module/hadoop-2.7.2/data/tmp/dfs/namesending incremental file list
    in_use.lock
    current/
    current/VERSION
    current/edits_0000000000000000001-0000000000000000008
    current/edits_0000000000000000009-0000000000000000010
    current/fsimage_0000000000000000008
    current/fsimage_0000000000000000008.md5
    current/fsimage_0000000000000000010
    current/fsimage_0000000000000000010.md5
    
    sent 2440 bytes  received 168 bytes  1738.67 bytes/sec

（4）重新启动namenode

[admin@hadoop16 namesecondary]$ sbin/hadoop-daemon.sh start namenode

或者：

[admin@hadoop14 hadoop-2.7.2]$ sbin/stop-dfs.sh 
    Stopping namenodes on [hadoop14]
    hadoop14: no namenode to stop
    hadoop15: stopping datanode
    hadoop14: no datanode to stop
    hadoop16: stopping datanode
    Stopping secondary namenodes [hadoop16]
    hadoop16: stopping secondarynamenode
    [admin@hadoop14 hadoop-2.7.2]$ sbin/start-dfs.sh 
    Starting namenodes on [hadoop14]
    hadoop14: starting namenode, logging to /opt/module/hadoop-2.7.2/logs/hadoop-admin-namenode-hadoop14.out
    hadoop15: starting datanode, logging to /opt/module/hadoop-2.7.2/logs/hadoop-admin-datanode-hadoop15.out
    hadoop14: starting datanode, logging to /opt/module/hadoop-2.7.2/logs/hadoop-admin-datanode-hadoop14.out
    hadoop16: starting datanode, logging to /opt/module/hadoop-2.7.2/logs/hadoop-admin-datanode-hadoop16.out
    Starting secondary namenodes [hadoop16]
    hadoop16: starting secondarynamenode, logging to /opt/module/hadoop-2.7.2/logs/hadoop-admin-secondarynamenode-hadoop16.out

### 3、nameNode故障处理二：使用-importCheckpoint选项启动namenode守护进程，从而将SecondaryNameNode中数据拷贝到namenode目录中 ###

（1）删除namenode存储的数据（/opt/module/hadoop-2.7.2/data/tmp/dfs/name）

（2）杀死NameNode进程

[admin@hadoop14 hadoop-2.7.2]$ jps 
    7447 Jps
    5545 NodeManager
    7213 DataNode
    7070 NameNode
    [admin@hadoop14 hadoop-2.7.2]$ kill -9 7070

（3）修改hdfs-site.xml文件

[admin@hadoop14 hadoop-2.7.2]$ sudo vim etc/hadoop/hdfs-site.xml

添加：

<property>
      <name>dfs.namenode.checkpoint.period</name>
      <value>120</value>
    </property>
    
    <property>
      <name>dfs.namenode.name.dir</name>
      <value>/opt/module/hadoop-2.7.2/data/tmp/dfs/name</value>
    </property>

（4）如果SecondaryNameNode不和Namenode在一个主机节点上，需要将SecondaryNameNode存储数据的目录（namesecondary ）拷贝到Namenode存储数据的平级目录（name）。

[admin@hadoop16 namesecondary]$ rsync -rvl /opt/module/hadoop-2.7.2/data/tmp/dfs/namesecondary admin@hadoop14:/opt/module/hadoop-2.7.2/data/tmp/dfs
    sending incremental file list
    namesecondary/
    namesecondary/in_use.lock
    namesecondary/current/
    namesecondary/current/VERSION
    namesecondary/current/edits_0000000000000000001-0000000000000000008
    namesecondary/current/edits_0000000000000000009-0000000000000000010
    namesecondary/current/edits_0000000000000000011-0000000000000000012
    namesecondary/current/fsimage_0000000000000000010
    namesecondary/current/fsimage_0000000000000000010.md5
    namesecondary/current/fsimage_0000000000000000012
    namesecondary/current/fsimage_0000000000000000012.md5
    
    sent 2589 bytes  received 191 bytes  5560.00 bytes/sec
    total size is 1820  speedup is 0.65

结果：

[admin@hadoop14 dfs]$ ll
    总用量 8
    drwx------. 3 admin admin 4096 10月 20 14:49 data
    drwxrwxr-x. 2 admin admin 4096 10月 20 16:45 name
    [admin@hadoop14 dfs]$ ll
    总用量 12
    drwx------. 3 admin admin 4096 10月 20 14:49 data
    drwxrwxr-x. 2 admin admin 4096 10月 20 16:45 name
    drwxrwxr-x. 3 admin admin 4096 10月 20 16:55 namesecondary

（5）导入检查点数据（等待一会ctrl+c结束掉）

这个操作会将将namesecondary目录的数据拷贝到name目录下

[admin@hadoop14 dfs]$ bin/hdfs namenode -importCheckpoint

（6）启动namenode

[admin@hadoop14 hadoop-2.7.2]$ sbin/hadoop-daemon.sh start namenode
    starting namenode, logging to /opt/module/hadoop-2.7.2/logs/hadoop-admin-namenode-hadoop14.out

（7）如果提示文件锁了，可以删除in\_use.lock

[admin@hadoop14 hadoop-2.7.2]$ rm -rf /opt/module/hadoop-2.7.2/data/tmp/dfs/namesecondary/in_use.lock

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25ld2JpZV85MDc0ODY4NTI_size_16_color_FFFFFF_t_70_pic_center_pic_center]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L25ld2JpZV85MDc0ODY4NTI_size_16_color_FFFFFF_t_70_pic_center_pic_center]: /images/20220505/4105cac4dcd243f2b6d633d1ca8fed77.png