Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Hadoop >> mail # dev >> Compile and deploy source code for Hadoop 1.0.4


Copy link to this message
-
Re: Compile and deploy source code for Hadoop 1.0.4
Hi Trupti,

Welcome! My responses inline.

On Sat, Feb 9, 2013 at 7:59 PM, Trupti Gaikwad <[EMAIL PROTECTED]> wrote:
> Hi,
>
> I want to work on release 1.0.4 source code. As per Hadoop
> wiki HowToContribute, I can download source code from trunk or from release
> 1.0.4 tag.

Although I do not know your goal here, note that the trunk is the best
place to do dev work if your goal is also to get your work accepted at
the end. We allow 1.x to continue receiving improvements but refuse
divergence in features compared to trunk and and the ongoing branch-2
releases. Just something to consider!

> 1. Source code from hadoop/common/trunk with revision 1397701 corresponding
> to release 1.0.4:
> I downloaded the source with svn revision 1397701 mentioned in release tag.
> My source code gets compiled, however tar file created by build does not
> contain start-mapred.sh file? It does contain start-yarn.sh. Even if source
> revision is old, why I am not getting start-mapred.sh. I really don't want
> to use resourcemanager or node manager to run my mapred job. How can I
> start jobtracker and tasktracker?

Unfortunately SVN revisions aren't exactly what you think they are.
What you need is to actually checkout a tag, not a revision. To get a
1.0.4 tag checked out from the Apache SVN repository, your command
could be:

$ svn checkout http://svn.apache.org/repos/asf/hadoop/common/tags/release-1.0.4/
hadoop-1.0.4
$ cd hadoop-1.0.4/

Likewise, if you want to work on the tip of the 1.x branch instead,
checkout the branch "branch-1":

$ svn checkout http://svn.apache.org/repos/asf/hadoop/common/branches/branch-1/
hadoop-1
$ cd hadoop-1/

> 2. Source code from tag release 1.0.4:
> Hadoop wiki also mentions that, If I want to work against any specific
> release then I will have to download release tag.
> I copied my code to src and tried to build it. However my code is not
> getting compiled because as I have developed it in above hadoop-common
> project. I am getting compilation error as there are inconsistencies in
> org.apache.hadoop.fs.FileSystem interface. Shall I develop my class by
> implementing interfaces provided in release 1.0.4?

You're attempting to build trunk (accidentally, in your case). See
above for getting proper 1.x code.

However, if you still wish to build trunk, whose build system is
different from the older 1.x system, some simple notes for building
trunk can be found here:
http://wiki.apache.org/hadoop/QwertyManiac/BuildingHadoopTrunk

> So
> 1. How to get all projects from hadoop-common?
> 2. what is correct way to compile and deploy any changes in core for
> release 1.0.4?

I believe I've answered both questions in the above inlines. Do feel
free to post any further questions you have!

--
Harsh J