HDFS-5865. Update OfflineImageViewer document. Contributed by Akira Ajisaka.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1590100 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Haohui Mai 2014-04-25 18:43:41 +00:00
parent a059eadbe9
commit 445b742354
2 changed files with 101 additions and 325 deletions

View File

@ -404,6 +404,8 @@ Release 2.5.0 - UNRELEASED
HDFS-6276. Remove unnecessary conditions and null check. (suresh)
HDFS-5865. Update OfflineImageViewer document. (Akira Ajisaka via wheat9)
Release 2.4.1 - UNRELEASED
INCOMPATIBLE CHANGES

View File

@ -23,56 +23,29 @@ Offline Image Viewer Guide
* Overview
The Offline Image Viewer is a tool to dump the contents of hdfs fsimage
files to human-readable formats in order to allow offline analysis and
examination of an Hadoop cluster's namespace. The tool is able to
process very large image files relatively quickly, converting them to
one of several output formats. The tool handles the layout formats that
were included with Hadoop versions 16 and up. If the tool is not able
to process an image file, it will exit cleanly. The Offline Image
Viewer does not require an Hadoop cluster to be running; it is entirely
offline in its operation.
files to a human-readable format and provide read-only WebHDFS API
in order to allow offline analysis and examination of an Hadoop cluster's
namespace. The tool is able to process very large image files relatively
quickly. The tool handles the layout formats that were included with Hadoop
versions 2.4 and up. If you want to handle older layout formats, you can
use the Offline Image Viewer of Hadoop 2.3.
If the tool is not able to process an image file, it will exit cleanly.
The Offline Image Viewer does not require a Hadoop cluster to be running;
it is entirely offline in its operation.
The Offline Image Viewer provides several output processors:
[[1]] Ls is the default output processor. It closely mimics the format of
the lsr command. It includes the same fields, in the same order, as
lsr : directory or file flag, permissions, replication, owner,
group, file size, modification date, and full path. Unlike the lsr
command, the root path is included. One important difference
between the output of the lsr command this processor, is that this
output is not sorted by directory name and contents. Rather, the
files are listed in the order in which they are stored in the
fsimage file. Therefore, it is not possible to directly compare the
output of the lsr command this this tool. The Ls processor uses
information contained within the Inode blocks to calculate file
sizes and ignores the -skipBlocks option.
[[1]] Web is the default output processor. It launches a HTTP server
that exposes read-only WebHDFS API. Users can investigate the namespace
interactively by using HTTP REST API.
[[2]] Indented provides a more complete view of the fsimage's contents,
including all of the information included in the image, such as
image version, generation stamp and inode- and block-specific
listings. This processor uses indentation to organize the output
into a hierarchal manner. The lsr format is suitable for easy human
comprehension.
[[3]] Delimited provides one file per line consisting of the path,
replication, modification time, access time, block size, number of
blocks, file size, namespace quota, diskspace quota, permissions,
username and group name. If run against an fsimage that does not
contain any of these fields, the field's column will be included,
but no data recorded. The default record delimiter is a tab, but
this may be changed via the -delimiter command line argument. This
processor is designed to create output that is easily analyzed by
other tools, such as {{{http://pig.apache.org}Apache Pig}}. See
the {{Analyzing Results}} section for further information on using
this processor to analyze the contents of fsimage files.
[[4]] XML creates an XML document of the fsimage and includes all of the
[[2]] XML creates an XML document of the fsimage and includes all of the
information within the fsimage, similar to the lsr processor. The
output of this processor is amenable to automated processing and
analysis with XML tools. Due to the verbosity of the XML syntax,
this processor will also generate the largest amount of output.
[[5]] FileDistribution is the tool for analyzing file sizes in the
[[3]] FileDistribution is the tool for analyzing file sizes in the
namespace image. In order to run the tool one should define a range
of integers [0, maxSize] by specifying maxSize and a step. The
range of integers is divided into segments of size step: [0, s[1],
@ -86,105 +59,93 @@ Offline Image Viewer Guide
* Usage
** Basic
** Web Processor
The simplest usage of the Offline Image Viewer is to provide just an
input and output file, via the -i and -o command-line switches:
Web processor launches a HTTP server which exposes read-only WebHDFS API.
Users can specify the address to listen by -addr option (default by
localhost:5978).
----
bash$ bin/hdfs oiv -i fsimage -o fsimage.txt
bash$ bin/hdfs oiv -i fsimage
14/04/07 13:25:14 INFO offlineImageViewer.WebImageViewer: WebImageViewer
started. Listening on /127.0.0.1:5978. Press Ctrl+C to stop the viewer.
----
This will create a file named fsimage.txt in the current directory
using the Ls output processor. For very large image files, this process
may take several minutes.
One can specify which output processor via the command-line switch -p.
For instance:
Users can access the viewer and get the information of the fsimage by
the following shell command:
----
bash$ bin/hdfs oiv -i fsimage -o fsimage.xml -p XML
bash$ bin/hdfs dfs -ls webhdfs://127.0.0.1:5978/
Found 2 items
drwxrwx--- - root supergroup 0 2014-03-26 20:16 webhdfs://127.0.0.1:5978/tmp
drwxr-xr-x - root supergroup 0 2014-03-31 14:08 webhdfs://127.0.0.1:5978/user
----
or
To get the information of all the files and directories, you can simply use
the following command:
----
bash$ bin/hdfs oiv -i fsimage -o fsimage.txt -p Indented
bash$ bin/hdfs dfs -ls -R webhdfs://127.0.0.1:5978/
----
This will run the tool using either the XML or Indented output
processor, respectively.
One command-line option worth considering is -skipBlocks, which
prevents the tool from explicitly enumerating all of the blocks that
make up a file in the namespace. This is useful for file systems that
have very large files. Enabling this option can significantly decrease
the size of the resulting output, as individual blocks are not
included. Note, however, that the Ls processor needs to enumerate the
blocks and so overrides this option.
Example
Consider the following contrived namespace:
Users can also get JSON formatted FileStatuses via HTTP REST API.
----
drwxr-xr-x - theuser supergroup 0 2009-03-16 21:17 /anotherDir
-rw-r--r-- 3 theuser supergroup 286631664 2009-03-16 21:15 /anotherDir/biggerfile
-rw-r--r-- 3 theuser supergroup 8754 2009-03-16 21:17 /anotherDir/smallFile
drwxr-xr-x - theuser supergroup 0 2009-03-16 21:11 /mapredsystem
drwxr-xr-x - theuser supergroup 0 2009-03-16 21:11 /mapredsystem/theuser
drwxr-xr-x - theuser supergroup 0 2009-03-16 21:11 /mapredsystem/theuser/mapredsystem
drwx-wx-wx - theuser supergroup 0 2009-03-16 21:11 /mapredsystem/theuser/mapredsystem/ip.redacted.com
drwxr-xr-x - theuser supergroup 0 2009-03-16 21:12 /one
drwxr-xr-x - theuser supergroup 0 2009-03-16 21:12 /one/two
drwxr-xr-x - theuser supergroup 0 2009-03-16 21:16 /user
drwxr-xr-x - theuser supergroup 0 2009-03-16 21:19 /user/theuser
bash$ curl -i http://127.0.0.1:5978/webhdfs/v1/?op=liststatus
HTTP/1.1 200 OK
Content-Type: application/json
Content-Length: 252
{"FileStatuses":{"FileStatus":[
{"fileId":16386,"accessTime":0,"replication":0,"owner":"theuser","length":0,"permission":"755","blockSize":0,"modificationTime":1392772497282,"type":"DIRECTORY","group":"supergroup","childrenNum":1,"pathSuffix":"user"}
]}}
----
Applying the Offline Image Processor against this file with default
options would result in the following output:
The Web processor now supports the following operations:
* {{{./WebHDFS.html#List_a_Directory}LISTSTATUS}}
* {{{./WebHDFS.html#Status_of_a_FileDirectory}GETFILESTATUS}}
* {{{./WebHDFS.html#Get_ACL_Status}GETACLSTATUS}}
** XML Processor
XML Processor is used to dump all the contents in the fsimage. Users can
specify input and output file via -i and -o command-line.
----
machine:hadoop-0.21.0-dev theuser$ bin/hdfs oiv -i fsimagedemo -o fsimage.txt
drwxr-xr-x - theuser supergroup 0 2009-03-16 14:16 /
drwxr-xr-x - theuser supergroup 0 2009-03-16 14:17 /anotherDir
drwxr-xr-x - theuser supergroup 0 2009-03-16 14:11 /mapredsystem
drwxr-xr-x - theuser supergroup 0 2009-03-16 14:12 /one
drwxr-xr-x - theuser supergroup 0 2009-03-16 14:16 /user
-rw-r--r-- 3 theuser supergroup 286631664 2009-03-16 14:15 /anotherDir/biggerfile
-rw-r--r-- 3 theuser supergroup 8754 2009-03-16 14:17 /anotherDir/smallFile
drwxr-xr-x - theuser supergroup 0 2009-03-16 14:11 /mapredsystem/theuser
drwxr-xr-x - theuser supergroup 0 2009-03-16 14:11 /mapredsystem/theuser/mapredsystem
drwx-wx-wx - theuser supergroup 0 2009-03-16 14:11 /mapredsystem/theuser/mapredsystem/ip.redacted.com
drwxr-xr-x - theuser supergroup 0 2009-03-16 14:12 /one/two
drwxr-xr-x - theuser supergroup 0 2009-03-16 14:19 /user/theuser
bash$ bin/hdfs oiv -p XML -i fsimage -o fsimage.xml
----
Similarly, applying the Indented processor would generate output that
begins with:
This will create a file named fsimage.xml contains all the information in
the fsimage. For very large image files, this process may take several
minutes.
Applying the Offline Image Viewer with XML processor would result in the
following output:
----
machine:hadoop-0.21.0-dev theuser$ bin/hdfs oiv -i fsimagedemo -p Indented -o fsimage.txt
FSImage
ImageVersion = -19
NamespaceID = 2109123098
GenerationStamp = 1003
INodes [NumInodes = 12]
Inode
INodePath =
Replication = 0
ModificationTime = 2009-03-16 14:16
AccessTime = 1969-12-31 16:00
BlockSize = 0
Blocks [NumBlocks = -1]
NSQuota = 2147483647
DSQuota = -1
Permissions
Username = theuser
GroupName = supergroup
PermString = rwxr-xr-x
<?xml version="1.0"?>
<fsimage>
<NameSection>
<genstampV1>1000</genstampV1>
<genstampV2>1002</genstampV2>
<genstampV1Limit>0</genstampV1Limit>
<lastAllocatedBlockId>1073741826</lastAllocatedBlockId>
<txid>37</txid>
</NameSection>
<INodeSection>
<lastInodeId>16400</lastInodeId>
<inode>
<id>16385</id>
<type>DIRECTORY</type>
<name></name>
<mtime>1392772497282</mtime>
<permission>theuser:supergroup:rwxr-xr-x</permission>
<nsquota>9223372036854775807</nsquota>
<dsquota>-1</dsquota>
</inode>
...remaining output omitted...
----
@ -193,30 +154,32 @@ Example
*-----------------------:-----------------------------------+
| <<Flag>> | <<Description>> |
*-----------------------:-----------------------------------+
| <<<-i>>>\|<<<--inputFile>>> <input file> | Specify the input fsimage file to
| | process. Required.
| <<<-i>>>\|<<<--inputFile>>> <input file> | Specify the input fsimage file
| | to process. Required.
*-----------------------:-----------------------------------+
| <<<-o>>>\|<<<--outputFile>>> <output file> | Specify the output filename, if the
| | specified output processor generates one. If the specified file already
| | exists, it is silently overwritten. Required.
| <<<-o>>>\|<<<--outputFile>>> <output file> | Specify the output filename,
| | if the specified output processor generates one. If
| | the specified file already exists, it is silently
| | overwritten. (output to stdout by default)
*-----------------------:-----------------------------------+
| <<<-p>>>\|<<<--processor>>> <processor> | Specify the image processor to apply
| | against the image file. Currently valid options are Ls (default), XML
| | and Indented..
| <<<-p>>>\|<<<--processor>>> <processor> | Specify the image processor to
| | apply against the image file. Currently valid options
| | are Web (default), XML and FileDistribution.
*-----------------------:-----------------------------------+
| <<<-skipBlocks>>> | Do not enumerate individual blocks within files. This may
| | save processing time and outfile file space on namespaces with very
| | large files. The Ls processor reads the blocks to correctly determine
| | file sizes and ignores this option.
| <<<-addr>>> <address> | Specify the address(host:port) to listen.
| | (localhost:5978 by default). This option is used with
| | Web processor.
*-----------------------:-----------------------------------+
| <<<-printToScreen>>> | Pipe output of processor to console as well as specified
| | file. On extremely large namespaces, this may increase processing time
| | by an order of magnitude.
| <<<-maxSize>>> <size> | Specify the range [0, maxSize] of file sizes to be
| | analyzed in bytes (128GB by default). This option is
| | used with FileDistribution processor.
*-----------------------:-----------------------------------+
| <<<-delimiter>>> <arg>| When used in conjunction with the Delimited processor,
| | replaces the default tab delimiter with the string specified by arg.
| <<<-step>>> <size> | Specify the granularity of the distribution in bytes
| | (2MB by default). This option is used with
| | FileDistribution processor.
*-----------------------:-----------------------------------+
| <<<-h>>>\|<<<--help>>>| Display the tool usage and help information and exit.
| <<<-h>>>\|<<<--help>>>| Display the tool usage and help information and
| | exit.
*-----------------------:-----------------------------------+
* Analyzing Results
@ -224,193 +187,4 @@ Example
The Offline Image Viewer makes it easy to gather large amounts of data
about the hdfs namespace. This information can then be used to explore
file system usage patterns or find specific files that match arbitrary
criteria, along with other types of namespace analysis. The Delimited
image processor in particular creates output that is amenable to
further processing by tools such as [38]Apache Pig. Pig provides a
particularly good choice for analyzing these data as it is able to deal
with the output generated from a small fsimage but also scales up to
consume data from extremely large file systems.
The Delimited image processor generates lines of text separated, by
default, by tabs and includes all of the fields that are common between
constructed files and files that were still under constructed when the
fsimage was generated. Examples scripts are provided demonstrating how
to use this output to accomplish three tasks: determine the number of
files each user has created on the file system, find files were created
but have not accessed, and find probable duplicates of large files by
comparing the size of each file.
Each of the following scripts assumes you have generated an output file
using the Delimited processor named foo and will be storing the results
of the Pig analysis in a file named results.
** Total Number of Files for Each User
This script processes each path within the namespace, groups them by
the file owner and determines the total number of files each user owns.
----
numFilesOfEachUser.pig:
-- This script determines the total number of files each user has in
-- the namespace. Its output is of the form:
-- username, totalNumFiles
-- Load all of the fields from the file
A = LOAD '$inputFile' USING PigStorage('\t') AS (path:chararray,
replication:int,
modTime:chararray,
accessTime:chararray,
blockSize:long,
numBlocks:int,
fileSize:long,
NamespaceQuota:int,
DiskspaceQuota:int,
perms:chararray,
username:chararray,
groupname:chararray);
-- Grab just the path and username
B = FOREACH A GENERATE path, username;
-- Generate the sum of the number of paths for each user
C = FOREACH (GROUP B BY username) GENERATE group, COUNT(B.path);
-- Save results
STORE C INTO '$outputFile';
----
This script can be run against pig with the following command:
----
bin/pig -x local -param inputFile=../foo -param outputFile=../results ../numFilesOfEachUser.pig
----
The output file's content will be similar to that below:
----
bart 1
lisa 16
homer 28
marge 2456
----
** Files That Have Never Been Accessed
This script finds files that were created but whose access times were
never changed, meaning they were never opened or viewed.
----
neverAccessed.pig:
-- This script generates a list of files that were created but never
-- accessed, based on their AccessTime
-- Load all of the fields from the file
A = LOAD '$inputFile' USING PigStorage('\t') AS (path:chararray,
replication:int,
modTime:chararray,
accessTime:chararray,
blockSize:long,
numBlocks:int,
fileSize:long,
NamespaceQuota:int,
DiskspaceQuota:int,
perms:chararray,
username:chararray,
groupname:chararray);
-- Grab just the path and last time the file was accessed
B = FOREACH A GENERATE path, accessTime;
-- Drop all the paths that don't have the default assigned last-access time
C = FILTER B BY accessTime == '1969-12-31 16:00';
-- Drop the accessTimes, since they're all the same
D = FOREACH C GENERATE path;
-- Save results
STORE D INTO '$outputFile';
----
This script can be run against pig with the following command and its
output file's content will be a list of files that were created but
never viewed afterwards.
----
bin/pig -x local -param inputFile=../foo -param outputFile=../results ../neverAccessed.pig
----
** Probable Duplicated Files Based on File Size
This script groups files together based on their size, drops any that
are of less than 100mb and returns a list of the file size, number of
files found and a tuple of the file paths. This can be used to find
likely duplicates within the filesystem namespace.
----
probableDuplicates.pig:
-- This script finds probable duplicate files greater than 100 MB by
-- grouping together files based on their byte size. Files of this size
-- with exactly the same number of bytes can be considered probable
-- duplicates, but should be checked further, either by comparing the
-- contents directly or by another proxy, such as a hash of the contents.
-- The scripts output is of the type:
-- fileSize numProbableDuplicates {(probableDup1), (probableDup2)}
-- Load all of the fields from the file
A = LOAD '$inputFile' USING PigStorage('\t') AS (path:chararray,
replication:int,
modTime:chararray,
accessTime:chararray,
blockSize:long,
numBlocks:int,
fileSize:long,
NamespaceQuota:int,
DiskspaceQuota:int,
perms:chararray,
username:chararray,
groupname:chararray);
-- Grab the pathname and filesize
B = FOREACH A generate path, fileSize;
-- Drop files smaller than 100 MB
C = FILTER B by fileSize > 100L * 1024L * 1024L;
-- Gather all the files of the same byte size
D = GROUP C by fileSize;
-- Generate path, num of duplicates, list of duplicates
E = FOREACH D generate group AS fileSize, COUNT(C) as numDupes, C.path AS files;
-- Drop all the files where there are only one of them
F = FILTER E by numDupes > 1L;
-- Sort by the size of the files
G = ORDER F by fileSize;
-- Save results
STORE G INTO '$outputFile';
----
This script can be run against pig with the following command:
----
bin/pig -x local -param inputFile=../foo -param outputFile=../results ../probableDuplicates.pig
----
The output file's content will be similar to that below:
----
1077288632 2 {(/user/tennant/work1/part-00501),(/user/tennant/work1/part-00993)}
1077288664 4 {(/user/tennant/work0/part-00567),(/user/tennant/work0/part-03980),(/user/tennant/work1/part-00725),(/user/eccelston/output/part-03395)}
1077288668 3 {(/user/tennant/work0/part-03705),(/user/tennant/work0/part-04242),(/user/tennant/work1/part-03839)}
1077288698 2 {(/user/tennant/work0/part-00435),(/user/eccelston/output/part-01382)}
1077288702 2 {(/user/tennant/work0/part-03864),(/user/eccelston/output/part-03234)}
----
Each line includes the file size in bytes that was found to be
duplicated, the number of duplicates found, and a list of the
duplicated paths. Files less than 100MB are ignored, providing a
reasonable likelihood that files of these exact sizes may be
duplicates.
criteria, along with other types of namespace analysis.