[jira] [Resolved] (IMPALA-3700) Introduce upper limit to hash table size or HashBuckets for Streaming agg

["Tim Armstrong (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/IMPALA-3700?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Tim Armstrong resolved IMPALA-3700.
-----------------------------------
    Resolution: Duplicate

This is just one case of IMPALA-3706. I don't think it's worth tracking separately for now.

> Introduce upper limit to hash table size or HashBuckets for Streaming agg
> -------------------------------------------------------------------------
>
>                 Key: IMPALA-3700
>                 URL: https://issues.apache.org/jira/browse/IMPALA-3700
>             Project: IMPALA
>          Issue Type: New Feature
>          Components: Backend
>    Affects Versions: Impala 2.7.0
>            Reporter: Mostafa Mokhtar
>            Priority: Minor
>              Labels: memory, supportability
>         Attachments: query67.sql.1.out, query67.sql.1.profile.out
>
>
> Several queries consume all the memory limit in Streaming agg leaving very little memory left for other operators. 
> Attached is the profile for TPC-DS Q67 which fails for the reason mentioned.
> An alternative solution is to extend the operator interface to respond to memory pressure by spilling and releasing some memory back to the memory/block manager.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)