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

Switch to Threaded View
Chukwa, mail # dev - ChukwaStorer


Copy link to this message
-
Re: ChukwaStorer
Bill Graham 2010-11-12, 16:49
I would think adding another constructor that takes these options in
the signature would be the best approach.
On Wed, Nov 10, 2010 at 6:31 PM, Corbin Hoenes <[EMAIL PROTECTED]> wrote:
> I'm using the chukwa-pig7 contrib in my pig scripts and noticed it trys to compress the chukwa output when I do a store.
>
>
> @Override
>  public void setStoreLocation(String location, Job job) throws IOException {
>    FileOutputFormat.setOutputPath(job, new Path(location));
>    FileOutputFormat.setCompressOutput(job, true);
>    FileOutputFormat.setOutputCompressorClass(job, DefaultCodec.class);
>    job.setOutputKeyClass(ChukwaRecordKey.class);
>    job.setOutputValueClass(ChukwaRecord.class);
>  }
>
> What is the best way to make this configurable? I was hoping I could do something like this:
>
> @Override
>  public void setStoreLocation(String location, Job job) throws IOException {
>    FileOutputFormat.setOutputPath(job, new Path(location));
>    FileOutputFormat.setCompressOutput(job, this.compress);
>    if(this.compress)
>      FileOutputFormat.setOutputCompressorClass(job, DefaultCodec.class);
>    job.setOutputKeyClass(ChukwaRecordKey.class);
>    job.setOutputValueClass(ChukwaRecord.class);
>  }
>
> How to make the codec class pluggable/configurable as well--for example if one wanted to use LZO?
>
>