You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@hive.apache.org by "Prasanth Jayachandran (JIRA)" <ji...@apache.org> on 2018/09/06 23:32:00 UTC

[jira] [Updated] (HIVE-20202) Add profiler endpoint to HS2 and LLAP

     [ https://issues.apache.org/jira/browse/HIVE-20202?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Prasanth Jayachandran updated HIVE-20202:
-----------------------------------------
    Attachment: HIVE-20202.7.patch

> Add profiler endpoint to HS2 and LLAP
> -------------------------------------
>
>                 Key: HIVE-20202
>                 URL: https://issues.apache.org/jira/browse/HIVE-20202
>             Project: Hive
>          Issue Type: New Feature
>    Affects Versions: 4.0.0, 3.2.0
>            Reporter: Prasanth Jayachandran
>            Assignee: Prasanth Jayachandran
>            Priority: Major
>         Attachments: HIVE-20202.1.patch, HIVE-20202.2.patch, HIVE-20202.3.patch, HIVE-20202.4.patch, HIVE-20202.5.patch, HIVE-20202.6.patch, HIVE-20202.7.patch
>
>
> Add a web endpoint for profiling based on async-profiler. This servlet should be added to httpserver so that HS2 and LLAP daemons can output flamegraphs when their /prof endpoint is hit. Since this will be based on [https://github.com/jvm-profiling-tools/async-profiler] heap allocation, lock contentions, HW counters etc. will also be supported in addition to cpu profiling. In most cases the profiling overhead is pretty low and is safe to run on production. More analysis on CPU and memory overhead here [https://github.com/jvm-profiling-tools/async-profiler/issues/14] and [https://github.com/jvm-profiling-tools/async-profiler/issues/131] 
>  
> For the impatient, here is the usage doc and the sample output [https://github.com/prasanthj/nightswatch/blob/master/README.md] 



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