That is a normal use case.
I'd encourage you to use Java MR (even pig/hive).
If you really want to use your legacy app use streaming with a map cmd
such as 'hadoop fs -cat <file> | mylegacyexe'
Sent from my iPhone
On Aug 23, 2011, at 8:00 AM, Zhixuan Zhu <[EMAIL PROTECTED]> wrote:
> I'll actually invoke one executable from each of my map. Because this
> C++ program has been implemented and used in the past, I just want to
> integrate it to our Hadoop map/reduce without having to re-implement the
> process in java. So my map is going to be very simple with just calling
> the process and pass the input files.
> -----Original Message-----
> From: Arun C Murthy [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, August 23, 2011 9:51 AM
> To: [EMAIL PROTECTED]
> Subject: Re: how to pass a hdfs file to a c++ process
> On Aug 22, 2011, at 12:57 PM, Zhixuan Zhu wrote:
>> Hi All,
>> I'm using hadoop-0.20.2 to try out some simple tasks. I asked a
>> about FileInputFormat a few days ago and get some prompt replys from
>> this forum and it helped a lot. Thanks again! Now I have another
>> question. I'm trying to invoke a C++ process from my mapper for each
>> hdfs file in the input directory to achieve some parallel processing.
> That seems weird - why aren't you using more maps and one file per-map?
>> But how do I pass the file to the program? I would want to do
>> like the following in my mapper:
> IAC, libhdfs is one way to do HDFS ops via c/c++.
>> Process lChldProc = Runtime.getRuntime().exec("myprocess -file
>> How do I pass the hdfs filesystem to an outside process like that? Is
>> HadoopStreaming the direction I should go?
>> Thanks very much for any reply in advance.