datanode上的文件介绍见附件:
hadoop块:
摘于hadoop权威指南! 留为备份
hadoop上默认块的大小为64M
当文件小于64M时,以一个block计算,在实际的文件系统中,仅占原始文件的大小,外加一个.meta文件
例如:
中等文件,40M
[root@ungeo12 current]#
[root@ungeo8 xliu]# /usr/local/hadoop/bin/hadoop dfs -put /home/xliu/hadoop-0.20.3-dev.tar.gz /
[root@ungeo8 xliu]# /usr/local/hadoop/bin/hadoop dfs -lsr /
-rw-r--r-- 3 root supergroup 40972760 2012-10-27 23:00 /hadoop-0.20.3-dev.tar.gz
[root@ungeo8 xliu]# hadoop fsck /hadoop-0.20.3-dev.tar.gz -blocks -locations -files
/hadoop-0.20.3-dev.tar.gz 40972760 bytes, 1 block(s): OK
0. blk_-6680210538719580844_1005 len=40972760 repl=3 [192.168.1.11:50010, 192.168.1.10:50010, 192.168.1.12:50010]
Status: HEALTHY
Total size: 40972760 B
Total dirs: 0
Total files: 1
Total blocks (validated): 1 (avg. block size 40972760 B)
Minimally replicated blocks: 1 (100.0 %)
Over-replicated blocks: 0 (0.0 %)
Under-replicated blocks: 0 (0.0 %)
Mis-replicated blocks: 0 (0.0 %)
Default replication factor: 3
Average block replication: 3.0
Corrupt blocks: 0
Missing replicas: 0 (0.0 %)
Number of data-nodes: 3
Number of racks: 1
The filesystem under path '/hadoop-0.20.3-dev.tar.gz' is HEALTHY
[root@ungeo8 xliu]# du -sh /home/xliu/hadoop-0.20.3-dev.tar.gz
40M /home/xliu/hadoop-0.20.3-dev.tar.gz ##在linux系统上原始的文件大小!
登陆到其中的一台datanode上查看相关的文件信息:
[root@ungeo10 current]# find . -name "*6680210538719580844*"
./blk_-6680210538719580844
./blk_-6680210538719580844_1005.meta
[root@ungeo10 current]# du -sh *6680210538719580844*
40M blk_-6680210538719580844
320K blk_-6680210538719580844_1005.meta
[root@ungeo10 current]#
大文件:81M
测试一个大于64M的文件
[root@ungeo8 xliu]# du -sh /home/xliu/jdk-6u21-linux-i586.bin
81M /home/xliu/jdk-6u21-linux-i586.bin
[root@ungeo8 xliu]# /usr/local/hadoop/bin/hadoop dfs -put /home/xliu/jdk-6u21-linux-i586.bin /
[root@ungeo8 xliu]# /usr/local/hadoop/bin/hadoop dfs -lsr /
-rw-r--r-- 3 root supergroup 40972760 2012-10-27 23:00 /hadoop-0.20.3-dev.tar.gz
-rw-r--r-- 3 root supergroup 83854743 2012-10-27 23:12 /jdk-6u21-linux-i586.bin
[root@ungeo8 xliu]# hadoop fsck /jdk-6u21-linux-i586.bin -blocks -locations -files;
/jdk-6u21-linux-i586.bin 83854743 bytes, 2 block(s): OK
0. blk_6729201486232919162_1006 len=67108864 repl=3 [192.168.1.10:50010, 192.168.1.11:50010, 192.168.1.12:50010]
1. blk_-2661147997145735854_1006 len=16745879 repl=3 [192.168.1.10:50010, 192.168.1.11:50010, 192.168.1.12:50010]
Status: HEALTHY
Total size: 83854743 B
Total dirs: 0
Total files: 1
Total blocks (validated): 2 (avg. block size 41927371 B)
Minimally replicated blocks: 2 (100.0 %)
Over-replicated blocks: 0 (0.0 %)
Under-replicated blocks: 0 (0.0 %)
Mis-replicated blocks: 0 (0.0 %)
Default replication factor: 3
Average block replication: 3.0
Corrupt blocks: 0
Missing replicas: 0 (0.0 %)
Number of data-nodes: 3
Number of racks: 1
The filesystem under path '/jdk-6u21-linux-i586.bin' is HEALTHY
[root@ungeo8 xliu]#
[root@ungeo12 ~]# cd /usr/local/hadoop
[root@ungeo12 hadoop]# ls
cAPACHE-README.txt CHANGES.txt hadoop-0.20.1-dev-ant.jar hadoop-0.20.1-dev-tools.jar LICENSE.txt tmp
bin conf hadoop-0.20.1-dev-core.jar ivy NOTICE.txt webapps
block contrib hadoop-0.20.1-dev-examples.jar ivy.xml README.txt YAHOO-CHANGES.txt
build.xml docs hadoop-0.20.1-dev-test.jar lib src
[root@ungeo12 hadoop]# cd block/
[root@ungeo12 block]# ls
blocksBeingWritten current detach in_use.lock storage tmp
[root@ungeo12 block]# cd current/
[root@ungeo12 current]# ls
blk_-2661147997145735854 blk_-6680210538719580844 blk_6729201486232919162 dncp_block_verification.log.curr
blk_-2661147997145735854_1006.meta blk_-6680210538719580844_1005.meta blk_6729201486232919162_1006.meta VERSION
[root@ungeo12 current]# pwd
/usr/local/hadoop/block/current
[root@ungeo12 current]# ls
blk_-2661147997145735854 blk_-6680210538719580844 blk_6729201486232919162 dncp_block_verification.log.curr
blk_-2661147997145735854_1006.meta blk_-6680210538719580844_1005.meta blk_6729201486232919162_1006.meta VERSION
[root@ungeo12 current]# find . -name "*6729201486232919162*"
./blk_6729201486232919162_1006.meta
./blk_6729201486232919162
[root@ungeo12 current]# du -sh *6729201486232919162*
65M blk_6729201486232919162 ##块1
520K blk_6729201486232919162_1006.meta
[root@ungeo12 current]#
[root@ungeo12 current]# du -sh *2661147997145735854*
16M blk_-2661147997145735854 ##块2
132K blk_-2661147997145735854_1006.meta
[root@ungeo12 current]#
hadoop块之间的复制!见附件
场景:
192.168.1.8 namenode
192.168.1.9 backupnode
192.168.1.10 datanode
192.168.1.11 datanode
192.168.1.12 datanode
用tail -f *.log抓图如下:
192.168.1.10 日志抓图:
192.168.1.12日志抓图:
192.168.1.11 日志抓图:
显然可以看出块在datanode直接进行复制,而且还有类似于ack回复!
当有数据块损坏时,namenode的处理方法:
阅读(2423) | 评论(0) | 转发(0) |