[
https://issues.apache.org/jira/browse/HIVE-25793?focusedWorklogId=695058&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-695058
]
ASF GitHub Bot logged work on HIVE-25793:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 13/Dec/21 13:44
Start Date: 13/Dec/21 13:44
Worklog Time Spent: 10m
Work Description: dengzhhu653 edited a comment on pull request #2860:
URL: https://github.com/apache/hive/pull/2860#issuecomment-992474538
> > I am not sure if it appropriate to introduce more metrics about the apis.
>
> I thought we may easily recognize retrying cases by comparing two api
metrics, whether apis succeeds or not. This approach doesn't break anything. It
gives us more accurate metrics, and users can pick metrics which they want to
collect, if they don't want both.
On a long running metastore instance, the two different metrics may be
almost the same, and some HMSHandler methods do not have `startFunction` and
`endFunction` to measure the time spent, it may bring some ambiguities and
memory pressure. Just my thought, Thank you!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 695058)
Time Spent: 1h 20m (was: 1h 10m)
> Isolate metastore metrics related to a retrying handler
> -------------------------------------------------------
>
> Key: HIVE-25793
> URL: https://issues.apache.org/jira/browse/HIVE-25793
> Project: Hive
> Issue Type: Bug
> Affects Versions: 2.3.8, 3.1.2, 4.0.0
> Reporter: Jeongdae Kim
> Assignee: Jeongdae Kim
> Priority: Minor
> Labels: pull-request-available
> Time Spent: 1h 20m
> Remaining Estimate: 0h
>
> We can see that a metastore api count increments twice for one RPC, because a
> metastore server makes two timers with same name for one RPC. one for a
> retrying proxy handler and another for a hms handler. It will make api
> metrics more inaccurate, especially during retrying.
> I think new metrics for retrying proxy will be helpful to make each metric
> accurate.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
