You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kudu.apache.org by "Todd Lipcon (JIRA)" <ji...@apache.org> on 2016/03/15 17:25:33 UTC

[jira] [Commented] (KUDU-1373) Publish table and tserver metrics in the master

    [ https://issues.apache.org/jira/browse/KUDU-1373?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15195583#comment-15195583 ] 

Todd Lipcon commented on KUDU-1373:
-----------------------------------

IMO we should be explicit in this feature not to "scope creep" into generalized metrics aggregation. Instead, we should only centralize and aggregate those metrics which are necessary for the master to make policy decisions (eg load balancing) or potentially for query engine optimizers (eg row counts or cardinality estimates). For the latter, it might make sense to consider distributing this info into an actual partitioned system table.

> Publish table and tserver metrics in the master
> -----------------------------------------------
>
>                 Key: KUDU-1373
>                 URL: https://issues.apache.org/jira/browse/KUDU-1373
>             Project: Kudu
>          Issue Type: Bug
>          Components: master, metrics, ops-tooling
>    Affects Versions: 0.7.1
>            Reporter: Adar Dembo
>
> This isn't fleshed out by any means, but some examples:
> * Table statistics (number of rows, perhaps?)
> * Tablet server load
> * Tablet load



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