[
https://issues.apache.org/jira/browse/HIVE-11220?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Gopal V updated HIVE-11220:
---------------------------
Attachment: hs2_Top_Components.zip
hs2_System_Overview.zip
hs2_Leak_Suspects.zip
> HS2 Lineage leakage with 16x concurrency tests
> ----------------------------------------------
>
> Key: HIVE-11220
> URL: https://issues.apache.org/jira/browse/HIVE-11220
> Project: Hive
> Issue Type: Bug
> Components: HiveServer2
> Affects Versions: llap, 1.3.0
> Reporter: Gopal V
> Attachments: hs2_Leak_Suspects.zip, hs2_System_Overview.zip,
> hs2_Top_Components.zip
>
>
> Test scenario is the HS2 + LLAP, 16x concurrency of TPCDS queries which take
> less than 4 seconds.
> session.LineageState accumulates optimizer lineage info and HS2 OOMs due to
> the amount of data being held in the SessionState, since the sessions are
> continously being used without pause.
> The issue seems to be triggered due to the volume of fast queries or the life
> time of a single JDBC connection.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
