Hadoop升级(No HA) 2.2升级到2.6

部署2.6.3

1 2 3 4 5 6 7 8 9 10 11 12 13 14

[hadoop@hadoop-master1 ~]$ tar zxvf hadoop-2.6.3.tar.gz [hadoop@hadoop-master1 ~]$ cd hadoop-2.6.3/share/ [hadoop@hadoop-master1 share]$ rm -rf doc/ [hadoop@hadoop-master1 hadoop-2.6.3]$ rm -rf lib/native/* #// 拷贝四个配置文件到hadoop-2.6.3 [hadoop@hadoop-master1 hadoop-2.6.3]$ cd etc/hadoop/ [hadoop@hadoop-master1 hadoop]$ cp -f ~/hadoop-2.2.0/etc/hadoop/*-site.xml ./ [hadoop@hadoop-master1 hadoop]$ cp -f ~/hadoop-2.2.0/etc/hadoop/slaves ./ [hadoop@hadoop-master1 hadoop]$ cd [hadoop@hadoop-master1 ~]$ for h in hadoop-master2 hadoop-slaver1 hadoop-slaver2 hadoop-slaver3 ; do rsync -vaz --delete --exclude=logs hadoop-2.6.3 $h:~/ ; done

升级（最佳方式）

直接使用upgrade选项启动dfs即可。（secondarynamenode不要单独操作来升级，反正就是执行upgrade启动dfs就好了）。

1 2 3 4 5

[hadoop@hadoop-master1 hadoop-2.6.3]$ sbin/start-dfs.sh -upgrade // 2.2和2.6都没有这个命令 // hadoop dfsadmin -upgradeProgress status hadoop dfsadmin -finalizeUpgrade

参考[Hadoop: The Definitive Guide/Chapter 10. Administering Hadoop/Maintenance/Upgrades]