crunch-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Gabriel Reid (JIRA)" <>
Subject [jira] [Updated] (CRUNCH-51) PCollection#sort relies on using a single reducer for total order sorting
Date Wed, 19 Sep 2012 09:00:07 GMT


Gabriel Reid updated CRUNCH-51:


Here's the test case(s) that I was using to detect issues with descending sorting and various
Avro variants (sorry for not posting this in the first place). The input file I was using
(words) is just a copy of the word dictionary in /usr/share/dict on my machine. Also, important
to note, I was running this using a pseudo-distributed cluster (i.e. not the local runner).

I didn't spot any clear errors when running into the crashes with Avro data, but the fact
that I'm getting a "Number of keys to write in partition file:0" seems pretty suspect to me.

If you'd like, I can take a stab at figuring out what's going on here later on today or tomorrow
as well.
> PCollection#sort relies on using a single reducer for total order sorting
> -------------------------------------------------------------------------
>                 Key: CRUNCH-51
>                 URL:
>             Project: Crunch
>          Issue Type: Improvement
>    Affects Versions: 0.3.0
>            Reporter: Gabriel Reid
>         Attachments: 0001-CRUNCH-51-Total-Order-Sort.patch, CRUNCH-51.patch, CRUNCH-51.patch,
> The total-order sorting provided by the Sort class (and therefore PCollection#sort) relies
on using a single reducer in order to provide total-order sorting. This is very inefficient
for large datasets, and should be replaced with a total order partitioner instead.
> For more information, see CRUNCH-23 (and possibly also MAPREDUCE-4574).

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

View raw message