[
https://issues.apache.org/jira/browse/PIG-1710?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Viraj Bhat updated PIG-1710:
----------------------------
Description:
We have a usecase in Pig where we cogroup on more than 2000 aliases.
{code}
cogroup_1 = foreach(cogroup A, B, C ... more than 2000 aliases ) generate
flatten (udf(....));
{code}
But we found out that there is a limitation on how many aliases a cogroup can
handle which is aound 127
So we had to workaround this by using 10-15 batches of 127 cogroups.
{code}
cogroup_1 = foreach(cogroup A, B, C ... 127aliases ) generate flatten
(udf(....));
cogroup_2 = foreach(cogroup A, B, C ... 127 aliases ) generate flatten
(udf(....));
...
cogroup_15 = foreach(cogroup A, B, C ... 127 aliases ) generate flatten
(udf(....));
{code}
Is there some documentation on this?
Viraj
was:
We have a usecase in Pig where we cogroup on more than 2000 aliases.
{code}
cogroup_1 = foreach(cogroup A, B, C ... more than 2000 aliases ) generate
flatten (udf(....));
{code}
But we found out that there is a limitation on how many aliases a cogroup can
handle which is 127 so we had to workaround in more than 10-15 of 127 batches
of 127 cogroups.
{code}
cogroup_1 = foreach(cogroup A, B, C ... 127aliases ) generate flatten
(udf(....));
cogroup_2 = foreach(cogroup A, B, C ... 127 aliases ) generate flatten
(udf(....));
...
cogroup_15 = foreach(cogroup A, B, C ... 127 aliases ) generate flatten
(udf(....));
{code}
Is there some documentation on this?
Viraj
> Document limitation on how many aliases cogroup can handle and why users
> should not cogroup more than 127 aliases
> -----------------------------------------------------------------------------------------------------------------
>
> Key: PIG-1710
> URL: https://issues.apache.org/jira/browse/PIG-1710
> Project: Pig
> Issue Type: Bug
> Components: documentation
> Reporter: Viraj Bhat
>
> We have a usecase in Pig where we cogroup on more than 2000 aliases.
> {code}
> cogroup_1 = foreach(cogroup A, B, C ... more than 2000 aliases ) generate
> flatten (udf(....));
> {code}
> But we found out that there is a limitation on how many aliases a cogroup can
> handle which is aound 127
> So we had to workaround this by using 10-15 batches of 127 cogroups.
> {code}
> cogroup_1 = foreach(cogroup A, B, C ... 127aliases ) generate flatten
> (udf(....));
> cogroup_2 = foreach(cogroup A, B, C ... 127 aliases ) generate flatten
> (udf(....));
> ...
> cogroup_15 = foreach(cogroup A, B, C ... 127 aliases ) generate flatten
> (udf(....));
> {code}
> Is there some documentation on this?
> Viraj
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
