MAPREDUCE-5639. Port DistCp2 document to trunk (Akira AJISAKA via jeagles)

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1590058 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Jonathan Turner Eagles 2014-04-25 15:23:42 +00:00
parent 5492149c3c
commit a059eadbe9
11 changed files with 495 additions and 862 deletions

View File

@ -167,6 +167,8 @@ Release 2.5.0 - UNRELEASED
MAPREDUCE-5852. Prepare MapReduce codebase for JUnit 4.11. (cnauroth)
MAPREDUCE-5639. Port DistCp2 document to trunk (Akira AJISAKA via jeagles)
OPTIMIZATIONS
BUG FIXES

View File

@ -0,0 +1,492 @@
<!---
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. See accompanying LICENSE file.
-->
#set ( $H3 = '###' )
DistCp Version2 Guide
=====================
---
- [Overview](#Overview)
- [Usage](#Usage)
- [Basic Usage](#Basic_Usage)
- [Update and Overwrite](#Update_and_Overwrite)
- [Command Line Options](#Command_Line_Options)
- [Architecture of DistCp](#Architecture_of_DistCp)
- [DistCp Driver](#DistCp_Driver)
- [Copy-listing Generator](#Copy-listing_Generator)
- [InputFormats and MapReduce Components](#InputFormats_and_MapReduce_Components)
- [Appendix](#Appendix)
- [Map sizing](#Map_sizing)
- [Copying Between Versions of HDFS](#Copying_Between_Versions_of_HDFS)
- [MapReduce and other side-effects](#MapReduce_and_other_side-effects)
- [SSL Configurations for HSFTP sources](#SSL_Configurations_for_HSFTP_sources)
- [Frequently Asked Questions](#Frequently_Asked_Questions)
---
Overview
--------
DistCp Version 2 (distributed copy) is a tool used for large
inter/intra-cluster copying. It uses MapReduce to effect its distribution,
error handling and recovery, and reporting. It expands a list of files and
directories into input to map tasks, each of which will copy a partition of
the files specified in the source list.
[The erstwhile implementation of DistCp]
(http://hadoop.apache.org/docs/r1.2.1/distcp.html) has its share of quirks
and drawbacks, both in its usage, as well as its extensibility and
performance. The purpose of the DistCp refactor was to fix these
shortcomings, enabling it to be used and extended programmatically. New
paradigms have been introduced to improve runtime and setup performance,
while simultaneously retaining the legacy behaviour as default.
This document aims to describe the design of the new DistCp, its spanking new
features, their optimal use, and any deviance from the legacy implementation.
Usage
-----
$H3 Basic Usage
The most common invocation of DistCp is an inter-cluster copy:
bash$ hadoop distcp hdfs://nn1:8020/foo/bar \
hdfs://nn2:8020/bar/foo
This will expand the namespace under `/foo/bar` on nn1 into a temporary file,
partition its contents among a set of map tasks, and start a copy on each
NodeManager from nn1 to nn2.
One can also specify multiple source directories on the command line:
bash$ hadoop distcp hdfs://nn1:8020/foo/a \
hdfs://nn1:8020/foo/b \
hdfs://nn2:8020/bar/foo
Or, equivalently, from a file using the -f option:
bash$ hadoop distcp -f hdfs://nn1:8020/srclist \
hdfs://nn2:8020/bar/foo
Where `srclist` contains
hdfs://nn1:8020/foo/a
hdfs://nn1:8020/foo/b
When copying from multiple sources, DistCp will abort the copy with an error
message if two sources collide, but collisions at the destination are
resolved per the [options](#Command_Line_Options) specified. By default,
files already existing at the destination are skipped (i.e. not replaced by
the source file). A count of skipped files is reported at the end of each
job, but it may be inaccurate if a copier failed for some subset of its
files, but succeeded on a later attempt.
It is important that each NodeManager can reach and communicate with both the
source and destination file systems. For HDFS, both the source and
destination must be running the same version of the protocol or use a
backwards-compatible protocol; see [Copying Between Versions]
(#Copying_Between_Versions_of_HDFS).
After a copy, it is recommended that one generates and cross-checks a listing
of the source and destination to verify that the copy was truly successful.
Since DistCp employs both Map/Reduce and the FileSystem API, issues in or
between any of the three could adversely and silently affect the copy. Some
have had success running with `-update` enabled to perform a second pass, but
users should be acquainted with its semantics before attempting this.
It's also worth noting that if another client is still writing to a source
file, the copy will likely fail. Attempting to overwrite a file being written
at the destination should also fail on HDFS. If a source file is (re)moved
before it is copied, the copy will fail with a FileNotFoundException.
Please refer to the detailed Command Line Reference for information on all
the options available in DistCp.
$H3 Update and Overwrite
`-update` is used to copy files from source that don't exist at the target,
or have different contents. `-overwrite` overwrites target-files even if they
exist at the source, or have the same contents.
Update and Overwrite options warrant special attention, since their handling
of source-paths varies from the defaults in a very subtle manner. Consider a
copy from `/source/first/` and `/source/second/` to `/target/`, where the
source paths have the following contents:
hdfs://nn1:8020/source/first/1
hdfs://nn1:8020/source/first/2
hdfs://nn1:8020/source/second/10
hdfs://nn1:8020/source/second/20
When DistCp is invoked without `-update` or `-overwrite`, the DistCp defaults
would create directories `first/` and `second/`, under `/target`. Thus:
distcp hdfs://nn1:8020/source/first hdfs://nn1:8020/source/second hdfs://nn2:8020/target
would yield the following contents in `/target`:
hdfs://nn2:8020/target/first/1
hdfs://nn2:8020/target/first/2
hdfs://nn2:8020/target/second/10
hdfs://nn2:8020/target/second/20
When either `-update` or `-overwrite` is specified, the **contents** of the
source-directories are copied to target, and not the source directories
themselves. Thus:
distcp -update hdfs://nn1:8020/source/first hdfs://nn1:8020/source/second hdfs://nn2:8020/target
would yield the following contents in `/target`:
hdfs://nn2:8020/target/1
hdfs://nn2:8020/target/2
hdfs://nn2:8020/target/10
hdfs://nn2:8020/target/20
By extension, if both source folders contained a file with the same name
(say, `0`), then both sources would map an entry to `/target/0` at the
destination. Rather than to permit this conflict, DistCp will abort.
Now, consider the following copy operation:
distcp hdfs://nn1:8020/source/first hdfs://nn1:8020/source/second hdfs://nn2:8020/target
With sources/sizes:
hdfs://nn1:8020/source/first/1 32
hdfs://nn1:8020/source/first/2 32
hdfs://nn1:8020/source/second/10 64
hdfs://nn1:8020/source/second/20 32
And destination/sizes:
hdfs://nn2:8020/target/1 32
hdfs://nn2:8020/target/10 32
hdfs://nn2:8020/target/20 64
Will effect:
hdfs://nn2:8020/target/1 32
hdfs://nn2:8020/target/2 32
hdfs://nn2:8020/target/10 64
hdfs://nn2:8020/target/20 32
`1` is skipped because the file-length and contents match. `2` is copied
because it doesn't exist at the target. `10` and `20` are overwritten since
the contents don't match the source.
If `-update` is used, `1` is overwritten as well.
Command Line Options
--------------------
Flag | Description | Notes
----------------- | ------------------------------------ | --------
`-p[rbugpca]` | Preserve r: replication number b: block size u: user g: group p: permission c: checksum-type a: ACL | Modification times are not preserved. Also, when `-update` is specified, status updates will **not** be synchronized unless the file sizes also differ (i.e. unless the file is re-created). If -pa is specified, DistCp preserves the permissions also because ACLs are a super-set of permissions.
`-i` | Ignore failures | As explained in the Appendix, this option will keep more accurate statistics about the copy than the default case. It also preserves logs from failed copies, which can be valuable for debugging. Finally, a failing map will not cause the job to fail before all splits are attempted.
`-log <logdir>` | Write logs to \<logdir\> | DistCp keeps logs of each file it attempts to copy as map output. If a map fails, the log output will not be retained if it is re-executed.
`-m <num_maps>` | Maximum number of simultaneous copies | Specify the number of maps to copy data. Note that more maps may not necessarily improve throughput.
`-overwrite` | Overwrite destination | If a map fails and `-i` is not specified, all the files in the split, not only those that failed, will be recopied. As discussed in the Usage documentation, it also changes the semantics for generating destination paths, so users should use this carefully.
`-update` | Overwrite if src size different from dst size | As noted in the preceding, this is not a "sync" operation. The only criterion examined is the source and destination file sizes; if they differ, the source file replaces the destination file. As discussed in the Usage documentation, it also changes the semantics for generating destination paths, so users should use this carefully.
`-f <urilist_uri>` | Use list at \<urilist_uri\> as src list | This is equivalent to listing each source on the command line. The `urilist_uri` list should be a fully qualified URI.
`-filelimit <n>` | Limit the total number of files to be <= n | **Deprecated!** Ignored in the new DistCp.
`-sizelimit <n>` | Limit the total size to be <= n bytes | **Deprecated!** Ignored in the new DistCp.
`-delete` | Delete the files existing in the dst but not in src | The deletion is done by FS Shell. So the trash will be used, if it is enable.
`-strategy {dynamic|uniformsize}` | Choose the copy-strategy to be used in DistCp. | By default, uniformsize is used. (i.e. Maps are balanced on the total size of files copied by each map. Similar to legacy.) If "dynamic" is specified, `DynamicInputFormat` is used instead. (This is described in the Architecture section, under InputFormats.)
`-bandwidth` | Specify bandwidth per map, in MB/second. | Each map will be restricted to consume only the specified bandwidth. This is not always exact. The map throttles back its bandwidth consumption during a copy, such that the **net** bandwidth used tends towards the specified value.
`-atomic {-tmp <tmp_dir>}` | Specify atomic commit, with optional tmp directory. | `-atomic` instructs DistCp to copy the source data to a temporary target location, and then move the temporary target to the final-location atomically. Data will either be available at final target in a complete and consistent form, or not at all. Optionally, `-tmp` may be used to specify the location of the tmp-target. If not specified, a default is chosen. **Note:** tmp_dir must be on the final target cluster.
`-mapredSslConf <ssl_conf_file>` | Specify SSL Config file, to be used with HSFTP source | When using the hsftp protocol with a source, the security- related properties may be specified in a config-file and passed to DistCp. \<ssl_conf_file\> needs to be in the classpath.
`-async` | Run DistCp asynchronously. Quits as soon as the Hadoop Job is launched. | The Hadoop Job-id is logged, for tracking.
Architecture of DistCp
----------------------
The components of the new DistCp may be classified into the following
categories:
* DistCp Driver
* Copy-listing generator
* Input-formats and Map-Reduce components
$H3 DistCp Driver
The DistCp Driver components are responsible for:
* Parsing the arguments passed to the DistCp command on the command-line,
via:
* OptionsParser, and
* DistCpOptionsSwitch
* Assembling the command arguments into an appropriate DistCpOptions object,
and initializing DistCp. These arguments include:
* Source-paths
* Target location
* Copy options (e.g. whether to update-copy, overwrite, which
file-attributes to preserve, etc.)
* Orchestrating the copy operation by:
* Invoking the copy-listing-generator to create the list of files to be
copied.
* Setting up and launching the Hadoop Map-Reduce Job to carry out the
copy.
* Based on the options, either returning a handle to the Hadoop MR Job
immediately, or waiting till completion.
The parser-elements are exercised only from the command-line (or if
DistCp::run() is invoked). The DistCp class may also be used
programmatically, by constructing the DistCpOptions object, and initializing
a DistCp object appropriately.
$H3 Copy-listing Generator
The copy-listing-generator classes are responsible for creating the list of
files/directories to be copied from source. They examine the contents of the
source-paths (files/directories, including wild-cards), and record all paths
that need copy into a SequenceFile, for consumption by the DistCp Hadoop
Job. The main classes in this module include:
1. CopyListing: The interface that should be implemented by any
copy-listing-generator implementation. Also provides the factory method by
which the concrete CopyListing implementation is chosen.
2. SimpleCopyListing: An implementation of CopyListing that accepts multiple
source paths (files/directories), and recursively lists all the individual
files and directories under each, for copy.
3. GlobbedCopyListing: Another implementation of CopyListing that expands
wild-cards in the source paths.
4. FileBasedCopyListing: An implementation of CopyListing that reads the
source-path list from a specified file.
Based on whether a source-file-list is specified in the DistCpOptions, the
source-listing is generated in one of the following ways:
1. If there's no source-file-list, the GlobbedCopyListing is used. All
wild-cards are expanded, and all the expansions are forwarded to the
SimpleCopyListing, which in turn constructs the listing (via recursive
descent of each path).
2. If a source-file-list is specified, the FileBasedCopyListing is used.
Source-paths are read from the specified file, and then forwarded to the
GlobbedCopyListing. The listing is then constructed as described above.
One may customize the method by which the copy-listing is constructed by
providing a custom implementation of the CopyListing interface. The behaviour
of DistCp differs here from the legacy DistCp, in how paths are considered
for copy.
The legacy implementation only lists those paths that must definitely be
copied on to target. E.g. if a file already exists at the target (and
`-overwrite` isn't specified), the file isn't even considered in the
MapReduce Copy Job. Determining this during setup (i.e. before the MapReduce
Job) involves file-size and checksum-comparisons that are potentially
time-consuming.
The new DistCp postpones such checks until the MapReduce Job, thus reducing
setup time. Performance is enhanced further since these checks are
parallelized across multiple maps.
$H3 InputFormats and MapReduce Components
The InputFormats and MapReduce components are responsible for the actual copy
of files and directories from the source to the destination path. The
listing-file created during copy-listing generation is consumed at this
point, when the copy is carried out. The classes of interest here include:
* **UniformSizeInputFormat:**
This implementation of org.apache.hadoop.mapreduce.InputFormat provides
equivalence with Legacy DistCp in balancing load across maps. The aim of
the UniformSizeInputFormat is to make each map copy roughly the same number
of bytes. Apropos, the listing file is split into groups of paths, such
that the sum of file-sizes in each InputSplit is nearly equal to every
other map. The splitting isn't always perfect, but its trivial
implementation keeps the setup-time low.
* **DynamicInputFormat and DynamicRecordReader:**
The DynamicInputFormat implements org.apache.hadoop.mapreduce.InputFormat,
and is new to DistCp. The listing-file is split into several "chunk-files",
the exact number of chunk-files being a multiple of the number of maps
requested for in the Hadoop Job. Each map task is "assigned" one of the
chunk-files (by renaming the chunk to the task's id), before the Job is
launched.
Paths are read from each chunk using the DynamicRecordReader, and
processed in the CopyMapper. After all the paths in a chunk are processed,
the current chunk is deleted and a new chunk is acquired. The process
continues until no more chunks are available.
This "dynamic" approach allows faster map-tasks to consume more paths than
slower ones, thus speeding up the DistCp job overall.
* **CopyMapper:**
This class implements the physical file-copy. The input-paths are checked
against the input-options (specified in the Job's Configuration), to
determine whether a file needs copy. A file will be copied only if at least
one of the following is true:
* A file with the same name doesn't exist at target.
* A file with the same name exists at target, but has a different file
size.
* A file with the same name exists at target, but has a different
checksum, and `-skipcrccheck` isn't mentioned.
* A file with the same name exists at target, but `-overwrite` is
specified.
* A file with the same name exists at target, but differs in block-size
(and block-size needs to be preserved.
* **CopyCommitter:** This class is responsible for the commit-phase of the
DistCp job, including:
* Preservation of directory-permissions (if specified in the options)
* Clean-up of temporary-files, work-directories, etc.
Appendix
--------
$H3 Map sizing
By default, DistCp makes an attempt to size each map comparably so that each
copies roughly the same number of bytes. Note that files are the finest level
of granularity, so increasing the number of simultaneous copiers (i.e. maps)
may not always increase the number of simultaneous copies nor the overall
throughput.
The new DistCp also provides a strategy to "dynamically" size maps, allowing
faster data-nodes to copy more bytes than slower nodes. Using `-strategy
dynamic` (explained in the Architecture), rather than to assign a fixed set
of source-files to each map-task, files are instead split into several sets.
The number of sets exceeds the number of maps, usually by a factor of 2-3.
Each map picks up and copies all files listed in a chunk. When a chunk is
exhausted, a new chunk is acquired and processed, until no more chunks
remain.
By not assigning a source-path to a fixed map, faster map-tasks (i.e.
data-nodes) are able to consume more chunks, and thus copy more data, than
slower nodes. While this distribution isn't uniform, it is fair with regard
to each mapper's capacity.
The dynamic-strategy is implemented by the DynamicInputFormat. It provides
superior performance under most conditions.
Tuning the number of maps to the size of the source and destination clusters,
the size of the copy, and the available bandwidth is recommended for
long-running and regularly run jobs.
$H3 Copying Between Versions of HDFS
For copying between two different versions of Hadoop, one will usually use
HftpFileSystem. This is a read-only FileSystem, so DistCp must be run on the
destination cluster (more specifically, on NodeManagers that can write to the
destination cluster). Each source is specified as
`hftp://<dfs.http.address>/<path>` (the default `dfs.http.address` is
`<namenode>:50070`).
$H3 MapReduce and other side-effects
As has been mentioned in the preceding, should a map fail to copy one of its
inputs, there will be several side-effects.
* Unless `-overwrite` is specified, files successfully copied by a previous
map on a re-execution will be marked as "skipped".
* If a map fails `mapreduce.map.maxattempts` times, the remaining map tasks
will be killed (unless `-i` is set).
* If `mapreduce.map.speculative` is set set final and true, the result of the
copy is undefined.
$H3 SSL Configurations for HSFTP sources
To use an HSFTP source (i.e. using the hsftp protocol), a SSL configuration
file needs to be specified (via the `-mapredSslConf` option). This must
specify 3 parameters:
* `ssl.client.truststore.location`: The local-filesystem location of the
trust-store file, containing the certificate for the NameNode.
* `ssl.client.truststore.type`: (Optional) The format of the trust-store
file.
* `ssl.client.truststore.password`: (Optional) Password for the trust-store
file.
The following is an example of the contents of the contents of a SSL
Configuration file:
<configuration>
<property>
<name>ssl.client.truststore.location</name>
<value>/work/keystore.jks</value>
<description>Truststore to be used by clients like distcp. Must be specified.</description>
</property>
<property>
<name>ssl.client.truststore.password</name>
<value>changeme</value>
<description>Optional. Default value is "".</description>
</property>
<property>
<name>ssl.client.truststore.type</name>
<value>jks</value>
<description>Optional. Default value is "jks".</description>
</property>
</configuration>
The SSL configuration file must be in the class-path of the DistCp program.
Frequently Asked Questions
--------------------------
1. **Why does -update not create the parent source-directory under a pre-existing target directory?**
The behaviour of `-update` and `-overwrite` is described in detail in the
Usage section of this document. In short, if either option is used with a
pre-existing destination directory, the **contents** of each source
directory is copied over, rather than the source-directory itself. This
behaviour is consistent with the legacy DistCp implementation as well.
2. **How does the new DistCp differ in semantics from the Legacy DistCp?**
* Files that are skipped during copy used to also have their
file-attributes (permissions, owner/group info, etc.) unchanged, when
copied with Legacy DistCp. These are now updated, even if the file-copy
is skipped.
* Empty root directories among the source-path inputs were not created at
the target, in Legacy DistCp. These are now created.
3. **Why does the new DistCp use more maps than legacy DistCp?**
Legacy DistCp works by figuring out what files need to be actually copied
to target before the copy-job is launched, and then launching as many maps
as required for copy. So if a majority of the files need to be skipped
(because they already exist, for example), fewer maps will be needed. As a
consequence, the time spent in setup (i.e. before the M/R job) is higher.
The new DistCp calculates only the contents of the source-paths. It
doesn't try to filter out what files can be skipped. That decision is put
off till the M/R job runs. This is much faster (vis-a-vis execution-time),
but the number of maps launched will be as specified in the `-m` option,
or 20 (default) if unspecified.
4. **Why does DistCp not run faster when more maps are specified?**
At present, the smallest unit of work for DistCp is a file. i.e., a file
is processed by only one map. Increasing the number of maps to a value
exceeding the number of files would yield no performance benefit. The
number of maps launched would equal the number of files.
5. **Why does DistCp run out of memory?**
If the number of individual files/directories being copied from the source
path(s) is extremely large (e.g. 1,000,000 paths), DistCp might run out of
memory while determining the list of paths for copy. This is not unique to
the new DistCp implementation.
To get around this, consider changing the `-Xmx` JVM heap-size parameters,
as follows:
bash$ export HADOOP_CLIENT_OPTS="-Xms64m -Xmx1024m"
bash$ hadoop distcp /source /target

View File

@ -92,6 +92,7 @@
<item name="Encrypted Shuffle" href="hadoop-mapreduce-client/hadoop-mapreduce-client-core/EncryptedShuffle.html"/>
<item name="Pluggable Shuffle/Sort" href="hadoop-mapreduce-client/hadoop-mapreduce-client-core/PluggableShuffleAndPluggableSort.html"/>
<item name="Distributed Cache Deploy" href="hadoop-mapreduce-client/hadoop-mapreduce-client-core/DistributedCacheDeploy.html"/>
<item name="DistCp" href="hadoop-mapreduce-client/hadoop-mapreduce-client-core/DistCp.html"/>
</menu>
<menu name="YARN" inherit="top">

View File

@ -192,21 +192,6 @@
</execution>
</executions>
</plugin>
<!-- Disable generation of pdf using maven-pdf-plugin until v.1.2 is released.
See Hadoop 8064 for details. -->
<!--plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-pdf-plugin</artifactId>
<executions>
<execution>
<id>pdf</id>
<phase>package</phase>
<goals>
<goal>pdf</goal>
</goals>
</execution>
</executions>
</plugin-->
</plugins>
</build>
</project>

View File

@ -1,98 +0,0 @@
<?xml version="1.0" encoding="ISO-8859-1" ?>
<!--
Licensed to the Apache Software Foundation (ASF) under one
or more contributor license agreements. See the NOTICE file
distributed with this work for additional information
regarding copyright ownership. The ASF licenses this file
to you under the Apache License, Version 2.0 (the
"License"); you may not use this file except in compliance
with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing,
software distributed under the License is distributed on an
"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
KIND, either express or implied. See the License for the
specific language governing permissions and limitations
under the License.
-->
<faqs xmlns="http://maven.apache.org/FML/1.0.1"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/FML/1.0.1 http://maven.apache.org/xsd/fml-1.0.1.xsd"
title="Frequently Asked Questions">
<part id="General">
<title>General</title>
<faq id="Update">
<question>Why does -update not create the parent source-directory under
a pre-existing target directory?</question>
<answer>The behaviour of <code>-update</code> and <code>-overwrite</code>
is described in detail in the Usage section of this document. In short,
if either option is used with a pre-existing destination directory, the
<strong>contents</strong> of each source directory is copied over, rather
than the source-directory itself.
This behaviour is consistent with the legacy DistCp implementation as well.
</answer>
</faq>
<faq id="Deviation">
<question>How does the new DistCp differ in semantics from the Legacy
DistCp?</question>
<answer>
<ul>
<li>Files that are skipped during copy used to also have their
file-attributes (permissions, owner/group info, etc.) unchanged,
when copied with Legacy DistCp. These are now updated, even if
the file-copy is skipped.</li>
<li>Empty root directories among the source-path inputs were not
created at the target, in Legacy DistCp. These are now created.</li>
</ul>
</answer>
</faq>
<faq id="nMaps">
<question>Why does the new DistCp use more maps than legacy DistCp?</question>
<answer>
<p>Legacy DistCp works by figuring out what files need to be actually
copied to target <strong>before</strong> the copy-job is launched, and then
launching as many maps as required for copy. So if a majority of the files
need to be skipped (because they already exist, for example), fewer maps
will be needed. As a consequence, the time spent in setup (i.e. before the
M/R job) is higher.</p>
<p>The new DistCp calculates only the contents of the source-paths. It
doesn't try to filter out what files can be skipped. That decision is put-
off till the M/R job runs. This is much faster (vis-a-vis execution-time),
but the number of maps launched will be as specified in the <code>-m</code>
option, or 20 (default) if unspecified.</p>
</answer>
</faq>
<faq id="more_maps">
<question>Why does DistCp not run faster when more maps are specified?</question>
<answer>
<p>At present, the smallest unit of work for DistCp is a file. i.e.,
a file is processed by only one map. Increasing the number of maps to
a value exceeding the number of files would yield no performance
benefit. The number of maps lauched would equal the number of files.</p>
</answer>
</faq>
<faq id="client_mem">
<question>Why does DistCp run out of memory?</question>
<answer>
<p>If the number of individual files/directories being copied from
the source path(s) is extremely large (e.g. 1,000,000 paths), DistCp might
run out of memory while determining the list of paths for copy. This is
not unique to the new DistCp implementation.</p>
<p>To get around this, consider changing the <code>-Xmx</code> JVM
heap-size parameters, as follows:</p>
<p><code>bash$ export HADOOP_CLIENT_OPTS="-Xms64m -Xmx1024m"</code></p>
<p><code>bash$ hadoop distcp /source /target</code></p>
</answer>
</faq>
</part>
</faqs>

View File

@ -1,47 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<!--
Licensed to the Apache Software Foundation (ASF) under one
or more contributor license agreements. See the NOTICE file
distributed with this work for additional information
regarding copyright ownership. The ASF licenses this file
to you under the Apache License, Version 2.0 (the
"License"); you may not use this file except in compliance
with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing,
software distributed under the License is distributed on an
"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
KIND, either express or implied. See the License for the
specific language governing permissions and limitations
under the License.
-->
<!-- START SNIPPET: docDescriptor -->
<document xmlns="http://maven.apache.org/DOCUMENT/1.0.1"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/DOCUMENT/1.0.1 http://maven.apache.org/xsd/document-1.0.1.xsd"
outputName="distcp">
<meta>
<title>${project.name}</title>
</meta>
<toc name="Table of Contents">
<item name="Introduction" ref="index.xml"/>
<item name="Usage" ref="usage.xml"/>
<item name="Command Line Reference" ref="cli.xml"/>
<item name="Architecture" ref="architecture.xml"/>
<item name="Appendix" ref="appendix.xml"/>
<item name="FAQ" ref="faq.fml"/>
</toc>
<cover>
<coverTitle>${project.name}</coverTitle>
<coverSubTitle>v. ${project.version}</coverSubTitle>
<coverType>User Guide</coverType>
<projectName>${project.name}</projectName>
<companyName>Apache Hadoop</companyName>
</cover>
</document>

View File

@ -1,140 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<!--
Copyright 2002-2004 The Apache Software Foundation
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<document xmlns="http://maven.apache.org/XDOC/2.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/XDOC/2.0 http://maven.apache.org/xsd/xdoc-2.0.xsd">
<head>
<title>Appendix</title>
</head>
<body>
<section name="Map sizing">
<p> By default, DistCp makes an attempt to size each map comparably so
that each copies roughly the same number of bytes. Note that files are the
finest level of granularity, so increasing the number of simultaneous
copiers (i.e. maps) may not always increase the number of
simultaneous copies nor the overall throughput.</p>
<p> The new DistCp also provides a strategy to "dynamically" size maps,
allowing faster data-nodes to copy more bytes than slower nodes. Using
<code>-strategy dynamic</code> (explained in the Architecture), rather
than to assign a fixed set of source-files to each map-task, files are
instead split into several sets. The number of sets exceeds the number of
maps, usually by a factor of 2-3. Each map picks up and copies all files
listed in a chunk. When a chunk is exhausted, a new chunk is acquired and
processed, until no more chunks remain.</p>
<p> By not assigning a source-path to a fixed map, faster map-tasks (i.e.
data-nodes) are able to consume more chunks, and thus copy more data,
than slower nodes. While this distribution isn't uniform, it is
<strong>fair</strong> with regard to each mapper's capacity.</p>
<p>The dynamic-strategy is implemented by the DynamicInputFormat. It
provides superior performance under most conditions. </p>
<p>Tuning the number of maps to the size of the source and
destination clusters, the size of the copy, and the available
bandwidth is recommended for long-running and regularly run jobs.</p>
</section>
<section name="Copying between versions of HDFS">
<p>For copying between two different versions of Hadoop, one will
usually use HftpFileSystem. This is a read-only FileSystem, so DistCp
must be run on the destination cluster (more specifically, on
TaskTrackers that can write to the destination cluster). Each source is
specified as <code>hftp://&lt;dfs.http.address&gt;/&lt;path&gt;</code>
(the default <code>dfs.http.address</code> is
&lt;namenode&gt;:50070).</p>
</section>
<section name="Map/Reduce and other side-effects">
<p>As has been mentioned in the preceding, should a map fail to copy
one of its inputs, there will be several side-effects.</p>
<ul>
<li>Unless <code>-overwrite</code> is specified, files successfully
copied by a previous map on a re-execution will be marked as
&quot;skipped&quot;.</li>
<li>If a map fails <code>mapred.map.max.attempts</code> times, the
remaining map tasks will be killed (unless <code>-i</code> is
set).</li>
<li>If <code>mapred.speculative.execution</code> is set set
<code>final</code> and <code>true</code>, the result of the copy is
undefined.</li>
</ul>
</section>
<section name="SSL Configurations for HSFTP sources:">
<p>To use an HSFTP source (i.e. using the hsftp protocol), a Map-Red SSL
configuration file needs to be specified (via the <code>-mapredSslConf</code>
option). This must specify 3 parameters:</p>
<ul>
<li><code>ssl.client.truststore.location</code>: The local-filesystem
location of the trust-store file, containing the certificate for
the namenode.</li>
<li><code>ssl.client.truststore.type</code>: (Optional) The format of
the trust-store file.</li>
<li><code>ssl.client.truststore.password</code>: (Optional) Password
for the trust-store file.</li>
</ul>
<p>The following is an example of the contents of the contents of
a Map-Red SSL Configuration file:</p>
<p> <br/> <code> &lt;configuration&gt; </code> </p>
<p> <br/> <code>&lt;property&gt; </code> </p>
<p> <code>&lt;name&gt;ssl.client.truststore.location&lt;/name&gt; </code> </p>
<p> <code>&lt;value&gt;/work/keystore.jks&lt;/value&gt; </code> </p>
<p> <code>&lt;description&gt;Truststore to be used by clients like distcp. Must be specified. &lt;/description&gt;</code> </p>
<p> <br/> <code>&lt;/property&gt; </code> </p>
<p><code> &lt;property&gt; </code> </p>
<p> <code>&lt;name&gt;ssl.client.truststore.password&lt;/name&gt; </code> </p>
<p> <code>&lt;value&gt;changeme&lt;/value&gt; </code> </p>
<p> <code>&lt;description&gt;Optional. Default value is "". &lt;/description&gt; </code> </p>
<p> <code>&lt;/property&gt; </code> </p>
<p> <br/> <code> &lt;property&gt; </code> </p>
<p> <code> &lt;name&gt;ssl.client.truststore.type&lt;/name&gt;</code> </p>
<p> <code> &lt;value&gt;jks&lt;/value&gt;</code> </p>
<p> <code> &lt;description&gt;Optional. Default value is "jks". &lt;/description&gt;</code> </p>
<p> <code> &lt;/property&gt; </code> </p>
<p> <code> <br/> &lt;/configuration&gt; </code> </p>
<p><br/>The SSL configuration file must be in the class-path of the
DistCp program.</p>
</section>
</body>
</document>

View File

@ -1,215 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<!--
Copyright 2002-2004 The Apache Software Foundation
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<document xmlns="http://maven.apache.org/XDOC/2.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/XDOC/2.0 http://maven.apache.org/xsd/xdoc-2.0.xsd">
<head>
<title>Architecture of DistCp</title>
</head>
<body>
<section name="Architecture">
<p>The components of the new DistCp may be classified into the following
categories: </p>
<ul>
<li>DistCp Driver</li>
<li>Copy-listing generator</li>
<li>Input-formats and Map-Reduce components</li>
</ul>
<subsection name="DistCp Driver">
<p>The DistCp Driver components are responsible for:</p>
<ul>
<li>Parsing the arguments passed to the DistCp command on the
command-line, via:
<ul>
<li>OptionsParser, and</li>
<li>DistCpOptionsSwitch</li>
</ul>
</li>
<li>Assembling the command arguments into an appropriate
DistCpOptions object, and initializing DistCp. These arguments
include:
<ul>
<li>Source-paths</li>
<li>Target location</li>
<li>Copy options (e.g. whether to update-copy, overwrite, which
file-attributes to preserve, etc.)</li>
</ul>
</li>
<li>Orchestrating the copy operation by:
<ul>
<li>Invoking the copy-listing-generator to create the list of
files to be copied.</li>
<li>Setting up and launching the Hadoop Map-Reduce Job to carry
out the copy.</li>
<li>Based on the options, either returning a handle to the
Hadoop MR Job immediately, or waiting till completion.</li>
</ul>
</li>
</ul>
<br/>
<p>The parser-elements are exercised only from the command-line (or if
DistCp::run() is invoked). The DistCp class may also be used
programmatically, by constructing the DistCpOptions object, and
initializing a DistCp object appropriately.</p>
</subsection>
<subsection name="Copy-listing generator">
<p>The copy-listing-generator classes are responsible for creating the
list of files/directories to be copied from source. They examine
the contents of the source-paths (files/directories, including
wild-cards), and record all paths that need copy into a sequence-
file, for consumption by the DistCp Hadoop Job. The main classes in
this module include:</p>
<ol>
<li>CopyListing: The interface that should be implemented by any
copy-listing-generator implementation. Also provides the factory
method by which the concrete CopyListing implementation is
chosen.</li>
<li>SimpleCopyListing: An implementation of CopyListing that accepts
multiple source paths (files/directories), and recursively lists
all the individual files and directories under each, for
copy.</li>
<li>GlobbedCopyListing: Another implementation of CopyListing that
expands wild-cards in the source paths.</li>
<li>FileBasedCopyListing: An implementation of CopyListing that
reads the source-path list from a specified file.</li>
</ol>
<p/>
<p>Based on whether a source-file-list is specified in the
DistCpOptions, the source-listing is generated in one of the
following ways:</p>
<ol>
<li>If there's no source-file-list, the GlobbedCopyListing is used.
All wild-cards are expanded, and all the expansions are
forwarded to the SimpleCopyListing, which in turn constructs the
listing (via recursive descent of each path). </li>
<li>If a source-file-list is specified, the FileBasedCopyListing is
used. Source-paths are read from the specified file, and then
forwarded to the GlobbedCopyListing. The listing is then
constructed as described above.</li>
</ol>
<br/>
<p>One may customize the method by which the copy-listing is
constructed by providing a custom implementation of the CopyListing
interface. The behaviour of DistCp differs here from the legacy
DistCp, in how paths are considered for copy. </p>
<p>The legacy implementation only lists those paths that must
definitely be copied on to target.
E.g. if a file already exists at the target (and -overwrite isn't
specified), the file isn't even considered in the Map-Reduce Copy
Job. Determining this during setup (i.e. before the Map-Reduce Job)
involves file-size and checksum-comparisons that are potentially
time-consuming.</p>
<p>The new DistCp postpones such checks until the Map-Reduce Job, thus
reducing setup time. Performance is enhanced further since these
checks are parallelized across multiple maps.</p>
</subsection>
<subsection name="Input-formats and Map-Reduce components">
<p> The Input-formats and Map-Reduce components are responsible for
the actual copy of files and directories from the source to the
destination path. The listing-file created during copy-listing
generation is consumed at this point, when the copy is carried
out. The classes of interest here include:</p>
<ul>
<li><strong>UniformSizeInputFormat:</strong> This implementation of
org.apache.hadoop.mapreduce.InputFormat provides equivalence
with Legacy DistCp in balancing load across maps.
The aim of the UniformSizeInputFormat is to make each map copy
roughly the same number of bytes. Apropos, the listing file is
split into groups of paths, such that the sum of file-sizes in
each InputSplit is nearly equal to every other map. The splitting
isn't always perfect, but its trivial implementation keeps the
setup-time low.</li>
<li><strong>DynamicInputFormat and DynamicRecordReader:</strong>
<p> The DynamicInputFormat implements org.apache.hadoop.mapreduce.InputFormat,
and is new to DistCp. The listing-file is split into several
"chunk-files", the exact number of chunk-files being a multiple
of the number of maps requested for in the Hadoop Job. Each map
task is "assigned" one of the chunk-files (by renaming the chunk
to the task's id), before the Job is launched.</p>
<p>Paths are read from each chunk using the DynamicRecordReader,
and processed in the CopyMapper. After all the paths in a chunk
are processed, the current chunk is deleted and a new chunk is
acquired. The process continues until no more chunks are
available.</p>
<p>This "dynamic" approach allows faster map-tasks to consume
more paths than slower ones, thus speeding up the DistCp job
overall. </p>
</li>
<li><strong>CopyMapper:</strong> This class implements the physical
file-copy. The input-paths are checked against the input-options
(specified in the Job's Configuration), to determine whether a
file needs copy. A file will be copied only if at least one of
the following is true:
<ul>
<li>A file with the same name doesn't exist at target.</li>
<li>A file with the same name exists at target, but has a
different file size.</li>
<li>A file with the same name exists at target, but has a
different checksum, and -skipcrccheck isn't mentioned.</li>
<li>A file with the same name exists at target, but -overwrite
is specified.</li>
<li>A file with the same name exists at target, but differs in
block-size (and block-size needs to be preserved.</li>
</ul>
</li>
<li><strong>CopyCommitter:</strong>
This class is responsible for the commit-phase of the DistCp
job, including:
<ul>
<li>Preservation of directory-permissions (if specified in the
options)</li>
<li>Clean-up of temporary-files, work-directories, etc.</li>
</ul>
</li>
</ul>
</subsection>
</section>
</body>
</document>

View File

@ -1,138 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<!--
Copyright 2002-2004 The Apache Software Foundation
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<document xmlns="http://maven.apache.org/XDOC/2.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/XDOC/2.0 http://maven.apache.org/xsd/xdoc-2.0.xsd">
<head>
<title>Command Line Options</title>
</head>
<body>
<section name="Options Index">
<table>
<tr><th> Flag </th><th> Description </th><th> Notes </th></tr>
<tr><td><code>-p[rbugp]</code></td>
<td>Preserve<br/>
r: replication number<br/>
b: block size<br/>
u: user<br/>
g: group<br/>
p: permission<br/></td>
<td>Modification times are not preserved. Also, when
<code>-update</code> is specified, status updates will
<strong>not</strong> be synchronized unless the file sizes
also differ (i.e. unless the file is re-created).
</td></tr>
<tr><td><code>-i</code></td>
<td>Ignore failures</td>
<td>As explained in the Appendix, this option
will keep more accurate statistics about the copy than the
default case. It also preserves logs from failed copies, which
can be valuable for debugging. Finally, a failing map will not
cause the job to fail before all splits are attempted.
</td></tr>
<tr><td><code>-log &lt;logdir&gt;</code></td>
<td>Write logs to &lt;logdir&gt;</td>
<td>DistCp keeps logs of each file it attempts to copy as map
output. If a map fails, the log output will not be retained if
it is re-executed.
</td></tr>
<tr><td><code>-m &lt;num_maps&gt;</code></td>
<td>Maximum number of simultaneous copies</td>
<td>Specify the number of maps to copy data. Note that more maps
may not necessarily improve throughput.
</td></tr>
<tr><td><code>-overwrite</code></td>
<td>Overwrite destination</td>
<td>If a map fails and <code>-i</code> is not specified, all the
files in the split, not only those that failed, will be recopied.
As discussed in the Usage documentation, it also changes
the semantics for generating destination paths, so users should
use this carefully.
</td></tr>
<tr><td><code>-update</code></td>
<td>Overwrite if src size different from dst size</td>
<td>As noted in the preceding, this is not a &quot;sync&quot;
operation. The only criterion examined is the source and
destination file sizes; if they differ, the source file
replaces the destination file. As discussed in the
Usage documentation, it also changes the semantics for
generating destination paths, so users should use this carefully.
</td></tr>
<tr><td><code>-f &lt;urilist_uri&gt;</code></td>
<td>Use list at &lt;urilist_uri&gt; as src list</td>
<td>This is equivalent to listing each source on the command
line. The <code>urilist_uri</code> list should be a fully
qualified URI.
</td></tr>
<tr><td><code>-filelimit &lt;n&gt;</code></td>
<td>Limit the total number of files to be &lt;= n</td>
<td><strong>Deprecated!</strong> Ignored in the new DistCp.
</td></tr>
<tr><td><code>-sizelimit &lt;n&gt;</code></td>
<td>Limit the total size to be &lt;= n bytes</td>
<td><strong>Deprecated!</strong> Ignored in the new DistCp.
</td></tr>
<tr><td><code>-delete</code></td>
<td>Delete the files existing in the dst but not in src</td>
<td>The deletion is done by FS Shell. So the trash will be used,
if it is enable.
</td></tr>
<tr><td><code>-strategy {dynamic|uniformsize}</code></td>
<td>Choose the copy-strategy to be used in DistCp.</td>
<td>By default, uniformsize is used. (i.e. Maps are balanced on the
total size of files copied by each map. Similar to legacy.)
If "dynamic" is specified, <code>DynamicInputFormat</code> is
used instead. (This is described in the Architecture section,
under InputFormats.)
</td></tr>
<tr><td><code>-bandwidth</code></td>
<td>Specify bandwidth per map, in MB/second.</td>
<td>Each map will be restricted to consume only the specified
bandwidth. This is not always exact. The map throttles back
its bandwidth consumption during a copy, such that the
<strong>net</strong> bandwidth used tends towards the
specified value.
</td></tr>
<tr><td><code>-atomic {-tmp &lt;tmp_dir&gt;}</code></td>
<td>Specify atomic commit, with optional tmp directory.</td>
<td><code>-atomic</code> instructs DistCp to copy the source
data to a temporary target location, and then move the
temporary target to the final-location atomically. Data will
either be available at final target in a complete and consistent
form, or not at all.
Optionally, <code>-tmp</code> may be used to specify the
location of the tmp-target. If not specified, a default is
chosen. <strong>Note:</strong> tmp_dir must be on the final
target cluster.
</td></tr>
<tr><td><code>-mapredSslConf &lt;ssl_conf_file&gt;</code></td>
<td>Specify SSL Config file, to be used with HSFTP source</td>
<td>When using the hsftp protocol with a source, the security-
related properties may be specified in a config-file and
passed to DistCp. &lt;ssl_conf_file&gt; needs to be in
the classpath.
</td></tr>
<tr><td><code>-async</code></td>
<td>Run DistCp asynchronously. Quits as soon as the Hadoop
Job is launched.</td>
<td>The Hadoop Job-id is logged, for tracking.
</td></tr>
</table>
</section>
</body>
</document>

View File

@ -1,47 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<!--
Copyright 2002-2004 The Apache Software Foundation
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<document xmlns="http://maven.apache.org/XDOC/2.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/XDOC/2.0 http://maven.apache.org/xsd/xdoc-2.0.xsd">
<head>
<title>DistCp</title>
</head>
<body>
<section name="Overview">
<p>
DistCp (distributed copy) is a tool used for large inter/intra-cluster
copying. It uses Map/Reduce to effect its distribution, error
handling and recovery, and reporting. It expands a list of files and
directories into input to map tasks, each of which will copy a partition
of the files specified in the source list.
</p>
<p>
The erstwhile implementation of DistCp has its share of quirks and
drawbacks, both in its usage, as well as its extensibility and
performance. The purpose of the DistCp refactor was to fix these shortcomings,
enabling it to be used and extended programmatically. New paradigms have
been introduced to improve runtime and setup performance, while simultaneously
retaining the legacy behaviour as default.
</p>
<p>
This document aims to describe the design of the new DistCp, its spanking
new features, their optimal use, and any deviance from the legacy
implementation.
</p>
</section>
</body>
</document>

View File

@ -1,162 +0,0 @@
<!--
Copyright 2002-2004 The Apache Software Foundation
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<document xmlns="http://maven.apache.org/XDOC/2.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/XDOC/2.0 http://maven.apache.org/xsd/xdoc-2.0.xsd">
<head>
<title>Usage </title>
</head>
<body>
<section name="Basic Usage">
<p>The most common invocation of DistCp is an inter-cluster copy:</p>
<p><code>bash$ hadoop jar hadoop-distcp.jar hdfs://nn1:8020/foo/bar \</code><br/>
<code> hdfs://nn2:8020/bar/foo</code></p>
<p>This will expand the namespace under <code>/foo/bar</code> on nn1
into a temporary file, partition its contents among a set of map
tasks, and start a copy on each TaskTracker from nn1 to nn2.</p>
<p>One can also specify multiple source directories on the command
line:</p>
<p><code>bash$ hadoop jar hadoop-distcp.jar hdfs://nn1:8020/foo/a \</code><br/>
<code> hdfs://nn1:8020/foo/b \</code><br/>
<code> hdfs://nn2:8020/bar/foo</code></p>
<p>Or, equivalently, from a file using the <code>-f</code> option:<br/>
<code>bash$ hadoop jar hadoop-distcp.jar -f hdfs://nn1:8020/srclist \</code><br/>
<code> hdfs://nn2:8020/bar/foo</code><br/></p>
<p>Where <code>srclist</code> contains<br/>
<code>hdfs://nn1:8020/foo/a</code><br/>
<code>hdfs://nn1:8020/foo/b</code></p>
<p>When copying from multiple sources, DistCp will abort the copy with
an error message if two sources collide, but collisions at the
destination are resolved per the <a href="#options">options</a>
specified. By default, files already existing at the destination are
skipped (i.e. not replaced by the source file). A count of skipped
files is reported at the end of each job, but it may be inaccurate if a
copier failed for some subset of its files, but succeeded on a later
attempt.</p>
<p>It is important that each TaskTracker can reach and communicate with
both the source and destination file systems. For HDFS, both the source
and destination must be running the same version of the protocol or use
a backwards-compatible protocol (see <a href="#cpver">Copying Between
Versions</a>).</p>
<p>After a copy, it is recommended that one generates and cross-checks
a listing of the source and destination to verify that the copy was
truly successful. Since DistCp employs both Map/Reduce and the
FileSystem API, issues in or between any of the three could adversely
and silently affect the copy. Some have had success running with
<code>-update</code> enabled to perform a second pass, but users should
be acquainted with its semantics before attempting this.</p>
<p>It's also worth noting that if another client is still writing to a
source file, the copy will likely fail. Attempting to overwrite a file
being written at the destination should also fail on HDFS. If a source
file is (re)moved before it is copied, the copy will fail with a
FileNotFoundException.</p>
<p>Please refer to the detailed Command Line Reference for information
on all the options available in DistCp.</p>
</section>
<section name="Update and Overwrite">
<p><code>-update</code> is used to copy files from source that don't
exist at the target, or have different contents. <code>-overwrite</code>
overwrites target-files even if they exist at the source, or have the
same contents.</p>
<p><br/>Update and Overwrite options warrant special attention, since their
handling of source-paths varies from the defaults in a very subtle manner.
Consider a copy from <code>/source/first/</code> and
<code>/source/second/</code> to <code>/target/</code>, where the source
paths have the following contents:</p>
<p><code>hdfs://nn1:8020/source/first/1</code><br/>
<code>hdfs://nn1:8020/source/first/2</code><br/>
<code>hdfs://nn1:8020/source/second/10</code><br/>
<code>hdfs://nn1:8020/source/second/20</code><br/></p>
<p><br/>When DistCp is invoked without <code>-update</code> or
<code>-overwrite</code>, the DistCp defaults would create directories
<code>first/</code> and <code>second/</code>, under <code>/target</code>.
Thus:<br/></p>
<p><code>distcp hdfs://nn1:8020/source/first hdfs://nn1:8020/source/second hdfs://nn2:8020/target</code></p>
<p><br/>would yield the following contents in <code>/target</code>: </p>
<p><code>hdfs://nn2:8020/target/first/1</code><br/>
<code>hdfs://nn2:8020/target/first/2</code><br/>
<code>hdfs://nn2:8020/target/second/10</code><br/>
<code>hdfs://nn2:8020/target/second/20</code><br/></p>
<p><br/>When either <code>-update</code> or <code>-overwrite</code> is
specified, the <strong>contents</strong> of the source-directories
are copied to target, and not the source directories themselves. Thus: </p>
<p><code>distcp -update hdfs://nn1:8020/source/first hdfs://nn1:8020/source/second hdfs://nn2:8020/target</code></p>
<p><br/>would yield the following contents in <code>/target</code>: </p>
<p><code>hdfs://nn2:8020/target/1</code><br/>
<code>hdfs://nn2:8020/target/2</code><br/>
<code>hdfs://nn2:8020/target/10</code><br/>
<code>hdfs://nn2:8020/target/20</code><br/></p>
<p><br/>By extension, if both source folders contained a file with the same
name (say, <code>0</code>), then both sources would map an entry to
<code>/target/0</code> at the destination. Rather than to permit this
conflict, DistCp will abort.</p>
<p><br/>Now, consider the following copy operation:</p>
<p><code>distcp hdfs://nn1:8020/source/first hdfs://nn1:8020/source/second hdfs://nn2:8020/target</code></p>
<p><br/>With sources/sizes:</p>
<p><code>hdfs://nn1:8020/source/first/1 32</code><br/>
<code>hdfs://nn1:8020/source/first/2 32</code><br/>
<code>hdfs://nn1:8020/source/second/10 64</code><br/>
<code>hdfs://nn1:8020/source/second/20 32</code><br/></p>
<p><br/>And destination/sizes:</p>
<p><code>hdfs://nn2:8020/target/1 32</code><br/>
<code>hdfs://nn2:8020/target/10 32</code><br/>
<code>hdfs://nn2:8020/target/20 64</code><br/></p>
<p><br/>Will effect: </p>
<p><code>hdfs://nn2:8020/target/1 32</code><br/>
<code>hdfs://nn2:8020/target/2 32</code><br/>
<code>hdfs://nn2:8020/target/10 64</code><br/>
<code>hdfs://nn2:8020/target/20 32</code><br/></p>
<p><br/><code>1</code> is skipped because the file-length and contents match.
<code>2</code> is copied because it doesn't exist at the target.
<code>10</code> and <code>20</code> are overwritten since the contents
don't match the source. </p>
<p>If <code>-update</code> is used, <code>1</code> is overwritten as well.</p>
</section>
</body>
</document>