[
https://issues.apache.org/jira/browse/HDFS-10778?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15466139#comment-15466139
]
Akira Ajisaka commented on HDFS-10778:
--------------------------------------
Thanks [~linyiqun] for updating the patch. I tried your patch and get the
following output:
{noformat}
[centos@ip-172-31-21-203 conf]$ hdfs oiv -p FileDistribution -format -step 30
-maxSize 300 -i /hadoop/dfs/name/current/fsimage_0000000000000000307
Processed 0 inodes.
Size Range NumFiles
(0 B, 30 B] 2
(270 B, 300 B] 32
totalFiles = 34
totalDirectories = 13
totalBlocks = 34
totalSpace = 288198
maxFileSize = 160321
{noformat}
Actually maxFileSize is 160321 but the output says the file is in {{(270 B, 300
B\]}}. Would you fix it to output (270B, maxFileSize]?
If a fsimage includes empty files, the output is as follows:
{noformat}
Size Range NumFiles
(0 B, 0 B] 1
(0 B, 30 B] 2
{noformat}
I'm thinking \[0 B, 0 B\] is better than (0 B, 0 B\].
> Optimize the output result of FileDistribution processor in hdfs oiv command
> ----------------------------------------------------------------------------
>
> Key: HDFS-10778
> URL: https://issues.apache.org/jira/browse/HDFS-10778
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: tools
> Affects Versions: 2.7.1
> Reporter: Yiqun Lin
> Assignee: Yiqun Lin
> Priority: Minor
> Attachments: HDFS-10778.001.patch, HDFS-10778.002.patch,
> HDFS-10778.003.patch, HDFS-10778.004.patch, HDFS-10778.005.patch
>
>
> Now It's not directly to understand the output result of the
> {{FileDistribution}} processor that in hdfs oiv command for users. For
> example, this is a original output:
> {code}
> Size NumFiles
> 0 22556
> 1048576 404971
> 2097152 29259
> 3145728 16937
> 4194304 9197
> 5242880 6889
> 6291456 4930
> 7340032 4070
> 8388608 299384
> 9437184 274623
> {code}
> Two aspects make that hard to understand for users.
> First, the size column just showed as the number in byte, it's not readable
> here. The better way is showed with a binary prefix.
> Second, the size column would be better to showed as a size range. It will
> let users know the value in {{NumFiles}} column was counted from A size to B
> size.
> The expected output result should be this:
> {code}
> Size Range NumFiles
> (0 B, 0 B] 1666332
> (0 B, 1 M] 778473
> (1 M, 2 M] 35125
> (2 M, 3 M] 13978
> (3 M, 4 M] 10158
> (4 M, 5 M] 6970
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
