[jira] [Commented] (HIVE-18378) Explain plan should show if a Map/Reduce Work is being cached
[
https://issues.apache.org/jira/browse/HIVE-18378?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16464430#comment-16464430
]
Sahil Takiar commented on HIVE-18378:
-
Hey [~kellyzly] are you still working on this?
> Explain plan should show if a Map/Reduce Work is being cached
> -
>
> Key: HIVE-18378
> URL: https://issues.apache.org/jira/browse/HIVE-18378
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
>Reporter: Sahil Takiar
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18378.1.patch
>
>
> It would be nice if the explain plan showed what {{MapWork}} / {{ReduceWork}}
> objects are being cached by Spark.
> The {{CombineEquivalentWorkResolver}} is the only code that triggers Spark
> cache-ing, so we should be able to modify it so that it displays if a work
> object will be cached or not.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18378) Explain plan should show if a Map/Reduce Work is being cached
[
https://issues.apache.org/jira/browse/HIVE-18378?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16355893#comment-16355893
]
Sahil Takiar commented on HIVE-18378:
-
[~kellyzly] yes feel free to take this up. I spent some time working on this a
few weeks ago, but hit a few issues. I attached a patch of what I had done so
far. Feel free to use that the basis of this work (or not).
The main issue I was hitting was that the {{CombineEquivalentWorkResolver}} was
being triggered in a lot of different queries. More places than I thought it
should be.
> Explain plan should show if a Map/Reduce Work is being cached
> -
>
> Key: HIVE-18378
> URL: https://issues.apache.org/jira/browse/HIVE-18378
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
>Reporter: Sahil Takiar
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18378.1.patch
>
>
> It would be nice if the explain plan showed what {{MapWork}} / {{ReduceWork}}
> objects are being cached by Spark.
> The {{CombineEquivalentWorkResolver}} is the only code that triggers Spark
> cache-ing, so we should be able to modify it so that it displays if a work
> object will be cached or not.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18378) Explain plan should show if a Map/Reduce Work is being cached
[
https://issues.apache.org/jira/browse/HIVE-18378?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16340647#comment-16340647
]
liyunzhang commented on HIVE-18378:
---
[~stakiar]: can you assign this jira to me ? I guess this feature is need by
HIVE-17486 and once HIVE-17486 is resolved, it is better to show whether the
work is cached in explain.
> Explain plan should show if a Map/Reduce Work is being cached
> -
>
> Key: HIVE-18378
> URL: https://issues.apache.org/jira/browse/HIVE-18378
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
>Reporter: Sahil Takiar
>Assignee: liyunzhang
>Priority: Major
>
> It would be nice if the explain plan showed what {{MapWork}} / {{ReduceWork}}
> objects are being cached by Spark.
> The {{CombineEquivalentWorkResolver}} is the only code that triggers Spark
> cache-ing, so we should be able to modify it so that it displays if a work
> object will be cached or not.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18378) Explain plan should show if a Map/Reduce Work is being cached
[
https://issues.apache.org/jira/browse/HIVE-18378?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16312093#comment-16312093
]
Sahil Takiar commented on HIVE-18378:
-
The {{CombineEquivalentWorkResolver}} should also print something in the log
everytime it decides to combine two work objets. Right now it doesn't print
anything at the INFO level.
> Explain plan should show if a Map/Reduce Work is being cached
> -
>
> Key: HIVE-18378
> URL: https://issues.apache.org/jira/browse/HIVE-18378
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
>Reporter: Sahil Takiar
>
> It would be nice if the explain plan showed what {{MapWork}} / {{ReduceWork}}
> objects are being cached by Spark.
> The {{CombineEquivalentWorkResolver}} is the only code that triggers Spark
> cache-ing, so we should be able to modify it so that it displays if a work
> object will be cached or not.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
