[
https://issues.apache.org/jira/browse/MAPREDUCE-2616?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Amar Kamat updated MAPREDUCE-2616:
----------------------------------
Description:
Currently, all the Gridmix input data files are located at
<gridmix-io-dir>/input (<gridmix-io-dir> is expected as a CLI parameter). When
compression emulation is enabled, Gridmix will check for compressed files
(based on suffixes) in the input folder. Gridmix will bail out if there are no
compressed input files. If the input folder consists of a mix of compressed and
uncompressed input files, then Gridmix will only use compressed input files for
all the jobs. Gridmix should smartly assign
1. uncompressed input files for jobs the don't need input decompression
2. compressed input files for jobs that need input decompression
was:Currently, all the Gridmix input data files are located at
<gridmix-io-dir>/input (<gridmix-io-dir> is expected as a CLI parameter). When
compression emulation is enabled, Gridmix will check for compressed files
(based on suffixes) in the input folder. Gridmix will bail out if there are no
compressed input files. If the input folder consists of a mix of compressed and
uncompressed input files, then Gridmix might end up using uncompressed files
resulting into no emulation.
Priority: Minor (was: Major)
Issue Type: Improvement (was: Bug)
Summary: [Gridmix] InputStriper should smartly switch between
compressed and uncompressed files based on the simulated job's input data
characteristics (was: [Gridmix] Input data compression emulation might not
work as expected with data reuse)
MAPREDUCE-2408 modified {{InputStriper}} to take care of the previously
reported issue. This is just an optimization.
> [Gridmix] InputStriper should smartly switch between compressed and
> uncompressed files based on the simulated job's input data characteristics
> ----------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-2616
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-2616
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Components: contrib/gridmix
> Affects Versions: 0.23.0
> Reporter: Amar Kamat
> Assignee: Amar Kamat
> Priority: Minor
> Labels: compression-emulation, gridmix
> Fix For: 0.23.0
>
>
> Currently, all the Gridmix input data files are located at
> <gridmix-io-dir>/input (<gridmix-io-dir> is expected as a CLI parameter).
> When compression emulation is enabled, Gridmix will check for compressed
> files (based on suffixes) in the input folder. Gridmix will bail out if there
> are no compressed input files. If the input folder consists of a mix of
> compressed and uncompressed input files, then Gridmix will only use
> compressed input files for all the jobs. Gridmix should smartly assign
> 1. uncompressed input files for jobs the don't need input decompression
> 2. compressed input files for jobs that need input decompression
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
