[ 
https://issues.apache.org/jira/browse/HIVE-28717?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

László Bodor updated HIVE-28717:
--------------------------------
    Description: 
Need to check the values captured by 
[PerfLogger|https://github.com/apache/hive/blob/master/common/src/java/org/apache/hadoop/hive/ql/log/PerfLogger.java]
 and maybe add them as counters in the query summary

A few things to consider:
1.  maybe too many counters, there might be a query-level config for "whether 
to include perflogger values to counters"
2. some events, e.g. `getSplits` happen multiple times (Map1, Map2, ...), in 
which case a separate counter would be needed for all of them
3. check whether perflogger entries make their way back to HS2, if they happen 
in the TezAM, again, `getSplits`

  was:
Need to check the values captured by 
[PerfLogger|https://github.com/apache/hive/blob/master/common/src/java/org/apache/hadoop/hive/ql/log/PerfLogger.java]
 and maybe add them as counters in the query summary




> Expose perfLogger values to the query summary
> ---------------------------------------------
>
>                 Key: HIVE-28717
>                 URL: https://issues.apache.org/jira/browse/HIVE-28717
>             Project: Hive
>          Issue Type: Sub-task
>            Reporter: László Bodor
>            Priority: Major
>
> Need to check the values captured by 
> [PerfLogger|https://github.com/apache/hive/blob/master/common/src/java/org/apache/hadoop/hive/ql/log/PerfLogger.java]
>  and maybe add them as counters in the query summary
> A few things to consider:
> 1.  maybe too many counters, there might be a query-level config for "whether 
> to include perflogger values to counters"
> 2. some events, e.g. `getSplits` happen multiple times (Map1, Map2, ...), in 
> which case a separate counter would be needed for all of them
> 3. check whether perflogger entries make their way back to HS2, if they 
> happen in the TezAM, again, `getSplits`



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to