You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@samza.apache.org by "Yi Pan (Data Infrastructure) (JIRA)" <ji...@apache.org> on 2015/08/05 00:46:05 UTC

[jira] [Commented] (SAMZA-738) Samza Timer based metrics does not have enough precision

    [ https://issues.apache.org/jira/browse/SAMZA-738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14654491#comment-14654491 ] 

Yi Pan (Data Infrastructure) commented on SAMZA-738:
----------------------------------------------------

Hi, [~a.pejakovic], thanks for picking this up. I left some comments on RB. Also, ideally, we should update the flush-ms in KafkaSystemProducerMetrics as well.

Thanks!

> Samza Timer based metrics does not have enough precision
> --------------------------------------------------------
>
>                 Key: SAMZA-738
>                 URL: https://issues.apache.org/jira/browse/SAMZA-738
>             Project: Samza
>          Issue Type: Bug
>            Reporter: Yi Pan (Data Infrastructure)
>            Assignee: Aleksandar Pejakovic
>              Labels: newbie
>         Attachments: SAMZA-738.0.patch, SAMZA-738.1.patch
>
>
> The current metrics like process_ms, choose_ms, etc. only works at millisecond level precision and results in erroneous reports only containing values greater than 1ms latency. This does not work when each container's actual throughput is at multiple K qps. The precision of those metrics should be at nano-second level.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)