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 2015/10/05 03:24:27 UTC

[jira] [Updated] (STORM-888) Nimbus Scaling: very large topologies

     [ https://issues.apache.org/jira/browse/STORM-888?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Rick Kellogg updated STORM-888:
-------------------------------
    Component/s: storm-core

> Nimbus Scaling: very large topologies
> -------------------------------------
>
>                 Key: STORM-888
>                 URL: https://issues.apache.org/jira/browse/STORM-888
>             Project: Apache Storm
>          Issue Type: Bug
>          Components: storm-core
>            Reporter: Robert Joseph Evans
>
> While doing load testing of a heartbeat server implementation we found that nimbus started to struggle when scheduling a very large highly connected topology.
> [~knusbaum] perhaps you can give some details on the topology you used, but it was three levels spout, and two levels of bolts.  The bolts were very highly connected with separate streams for each connection.
> We should profile nimbus while it is doing this and see if we can determine the bottleneck to fix it.



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