You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@storm.apache.org by "Rick Kellogg (JIRA)" <ji...@apache.org> on 2015/10/09 02:29:31 UTC

[jira] [Updated] (STORM-141) Provide support for custom stats in Storm UI

     [ https://issues.apache.org/jira/browse/STORM-141?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Rick Kellogg updated STORM-141:
-------------------------------
    Component/s: storm-core

> Provide support for custom stats in Storm UI
> --------------------------------------------
>
>                 Key: STORM-141
>                 URL: https://issues.apache.org/jira/browse/STORM-141
>             Project: Apache Storm
>          Issue Type: New Feature
>          Components: storm-core
>            Reporter: James Xu
>            Priority: Minor
>
> https://github.com/nathanmarz/storm/issues/27
> Something along the lines of Hadoop counters
> ----------
> gsilk: Would this be implemented as an implicit stream? It would be great to have global and per-task aggregations in addition to having per-bolt counters.
> ----------
> nathanmarz: gsilk: Well, besides the aggregation mechanism we need to figure out how the stats are sent back to nimbus, stored, etc. Not sure the current approach of putting the stats in ZK is a good one. Using implicit streams and an aggregation bolt to sum things together, and then having the aggregation bolt connect to nimbus and send updates might be a better design.
> ----------
> ehudl: how about using emit to a specific stream (not default) as a counter, do you think this is a good approach ? 
> could this approach cause a memory leak? if no one is listening to this stream or in case you use a large amount of streams?



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