I do not think that anyone relies on part names produced by reducer -
NONE.
On Jun 11, 2007, at 7:59 AM, Arun C Murthy (JIRA) wrote:
[ https://issues.apache.org/jira/browse/HADOOP-1320?
page=com.atlassian.jira.plugin.system.issuetabpanels:comment-
tabpanel#action_12503427 ]
Arun C Murthy edited comment on HADOOP-1320 at 6/11/07 7:58 AM:
----------------------------------------------------------------
I realise, rather belatedly, that changing the filename of the
output of {{MapTask.DirectMapOutputCollector}} to make it
compatible with that of {{ReduceTask}} results in changes to output
of streaming's *-reducer NONE* too (e.g. I had to fix
{{org.apache.hadoop.streaming.TestStreamReduceNone}}) - is this a
big concern for streaming users? Runping/Arkady?
was:
I realise, rather belatedly, that changing the filename of the
output of {{MapTask.DirectMapOutputCollector}} to make it
compatible with that of {{ReduceTask}} results in changes to output
of streaming's *-reducer NONE* too (e.g. I had to fix
{{org.apache.hadoop.streaming.TestStreamReduceNone}}) - is this a
big concern for streaming users? Runping? Arkday?
Rewrite 'random-writer' to use '-reducer NONE'
----------------------------------------------
Key: HADOOP-1320
URL: https://issues.apache.org/jira/browse/
HADOOP-1320
Project: Hadoop
Issue Type: Improvement
Components: examples
Affects Versions: 0.12.3
Reporter: Arun C Murthy
Assignee: Arun C Murthy
Priority: Minor
Fix For: 0.14.0
Attachments: HADOOP-1320_1_20070611.patch,
HADOOP-1320_2_20070611.patch
Post HADOOP-1216 the 'random-writer' is archaic in the sense that
it uses custom code to write data to dfs, this is now supported by
the framework and doesn't make sense to reinvent the wheel here.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
