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

Switch to Threaded View
MapReduce, mail # user - RE: hadoop 1.0.3 equivalent of MultipleTextOutputFormat


Copy link to this message
-
RE: hadoop 1.0.3 equivalent of MultipleTextOutputFormat
Tony Burton 2012-08-31, 09:02
Hi Harsh,

I tried using NullOutputFormat as you suggested, however simply using

job.setOutputFormatClass(NullOutputFormat.class);

resulted in no output at all. Although I've not tried overriding getOutputCommitter in NullOutputFormat as you suggested, I discovered LazyOutputFormat which only writes when it has to, "the output file is created only when the first record is emitted for a given partition" (from "Hadoop: The Definitive Guide").

Instead of

job.setOutputFormatClass(TextOutputFormat.class);

use LazyOutputFormat like this:

LazyOutputFormat.setOutputFormatClass(job, TextOutputFormat.class);

So now my unnamed MultipleOutputs are handling to segmented results, and LazyOutputFormat is suppressing the default output. Good job!

Tony

________________________________________
From: Harsh J [[EMAIL PROTECTED]]
Sent: 29 August 2012 17:05
To: [EMAIL PROTECTED]
Subject: Re: hadoop 1.0.3 equivalent of MultipleTextOutputFormat

Hi Tony,

On Wed, Aug 29, 2012 at 9:30 PM, Tony Burton <[EMAIL PROTECTED]> wrote:
> Success so far!
>
> I followed the example given by Tom on the link to the MultipleOutputs.html API you suggested.
>
> I implemented a WordCount MR job using hadoop 1.0.3 and segmented the output depending on word length: output to directory "sml" for less than 10 characters, "med" for between 10 and 20 characters, "lrg" otherwise.
>
> I used out.write(key, new IntWritable(sum), generateFilename(key, sum)); to write the output, and generateFileName to create the custom directory name/filename. You need to provide the start of the filename as well otherwise your output files will be -r-00000, -r-00001 etc. (so, for example, return "sml/part"; etc)

Thanks for these notes, should come helpful for those who search!

> Also required: as Tom states, override Reducer.setup() to create the MultipleOutputs. However, Tom's puzzle left for the reader is that you also need to override Reducer.cleanup() and call close() on your MultipleOutputs object. Forget to do this and your segmented files will be empty.

Ah yes this is important. Non closure of files would have you wait for
an hour for data to get available to readers (open writer lease expiry
period).

> One observation: although it's not the end of the world, as well as my segmented output I also get a zero-size part-r-00000 file in the base of my output path. Is there any way to prevent creation of this file?

Set the OutputFormat to NullOutputFormat.

In case you face issues doing this in new API (you may notice some odd
behavior) try to extend NullOutputFormat and in its getOutputCommitter
method i.e. http://hadoop.apache.org/common/docs/r1.0.3/api/org/apache/hadoop/mapreduce/lib/output/NullOutputFormat.html#getOutputCommitter(org.apache.hadoop.mapreduce.TaskAttemptContext),
return a FileOutputCommitter object. By default it returns a no-op
OutputCommitter that may not gel well with a file-based writer such as
MultipleOutputs. Then set this new OutputFormat as your job's output
format.

> Thanks again Harsh for pointing the way.
>
> Tony
>
>
>
>
>
>
>
> -----Original Message-----
> From: Tony Burton [mailto:[EMAIL PROTECTED]]
> Sent: 29 August 2012 11:38
> To: [EMAIL PROTECTED]
> Subject: RE: hadoop 1.0.3 equivalent of MultipleTextOutputFormat
>
> Thanks Harsh! Will try it out and report back later.
>
>
> -----Original Message-----
> From: Harsh J [mailto:[EMAIL PROTECTED]]
> Sent: 29 August 2012 11:12
> To: [EMAIL PROTECTED]
> Subject: Re: hadoop 1.0.3 equivalent of MultipleTextOutputFormat
>
> Hi Tony,
>
> Seeing your new question, I recalled Tom's post to a user once, here:
> https://groups.google.com/a/cloudera.org/d/msg/cdh-user/pdyVyydt5Ys/1CaLukt4v1AJ
>
> This specific call allows you to specify / characters in your name,
> that gets translated into creation of directories automatically:
> http://hadoop.apache.org/common/docs/stable/api/org/apache/hadoop/mapreduce/lib/output/MultipleOutputs.html#write(KEYOUT,%20VALUEOUT,%20java.lang.String)

Harsh J
www.sportingindex.com
Inbound Email has been scanned for viruses and SPAM
**********************************************************************

This email and any attachments are confidential, protected by copyright and may be legally privileged.  If you are not the intended recipient, then the dissemination or copying of this email is prohibited. If you have received this in error, please notify the sender by replying by email and then delete the email completely from your system.  Neither Sporting Index nor the sender accepts responsibility for any virus, or any other defect which might affect any computer or IT system into which the email is received and/or opened.  It is the responsibility of the recipient to scan the email and no responsibility is accepted for any loss or damage arising in any way from receipt or use of this email.  Sporting Index Ltd is a company registered in England and Wales with company number 2636842, whose registered office is at Gateway House, Milverton Street, London, SE11 4AP.  Sporting Index Ltd is authorised and regulated by the UK Financial Services Authority (reg. no. 150404) and Gambling Commission (reg. no. 000-027343-R-308898-001).  Any financial promotion contained herein has been issued
and approved by Sporting Index Ltd.

Outbound email has been scanned for viruses and SPAM