[ 
https://issues.apache.org/jira/browse/HIVE-18936?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Sergey Shelukhin updated HIVE-18936:
------------------------------------
    Description: 
Due to Hive splits limitations (the only thing they have is file paths) the 
database/table/partition tags for cache entries (see HIVE-16992) are guesswork 
derived from file paths.
We should improve them; in some cases where split code is in Hive (e.g. 
OrcSplit) we can add information directly to the split.
For others, we can consider passing information as part of JobConf or in some 
other way. Hive execution already has path to partition mapping, so we may be 
able to utilize that.

  was:
Due to splits limitations the database/table/partition tags for cache entries 
(see HIVE-16992) are guesswork derived from file paths.
We should improve them; in some cases where split code is in Hive (e.g. 
OrcSplit) we can add information directly to the split.
For others, we can consider passing information as part of JobConf or in some 
other way.


> LLAP: improve data tags for cache view
> --------------------------------------
>
>                 Key: HIVE-18936
>                 URL: https://issues.apache.org/jira/browse/HIVE-18936
>             Project: Hive
>          Issue Type: Bug
>            Reporter: Sergey Shelukhin
>            Priority: Major
>
> Due to Hive splits limitations (the only thing they have is file paths) the 
> database/table/partition tags for cache entries (see HIVE-16992) are 
> guesswork derived from file paths.
> We should improve them; in some cases where split code is in Hive (e.g. 
> OrcSplit) we can add information directly to the split.
> For others, we can consider passing information as part of JobConf or in some 
> other way. Hive execution already has path to partition mapping, so we may be 
> able to utilize that.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to