[ 
https://issues.apache.org/jira/browse/STORM-1190?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Robert Joseph Evans resolved STORM-1190.
----------------------------------------
       Resolution: Fixed
         Assignee: Robert Joseph Evans
    Fix Version/s: 0.11.0

Thanks for all of the help from everyone on this.  Especially Daniel Schonfeld 
who proposed the final solution.  The new CPU utilization should be just 
slightly higher then it was without batching.

On a happy side note this reduced the CPU utilization everywhere with batching 
so now with all of the performance changes that have gone in I can run a 2 
worker ThroughputVsLatency topology at 43,000 sentences per second on my MBP, 
with a max spout pending of 100 and Automatic Back Pressure off.  I don't have 
an apples to apples comparison with before this change.  Because it looks like 
ABP was limiting things to about 35,000 sentences per second.  Either way we 
are running a lot better then we were before where we would max out at about 
6,500 sentences per second.

> System load spikes in recent snapshot
> -------------------------------------
>
>                 Key: STORM-1190
>                 URL: https://issues.apache.org/jira/browse/STORM-1190
>             Project: Apache Storm
>          Issue Type: Bug
>          Components: storm-core
>    Affects Versions: 0.11.0
>         Environment: 10x (CoreOS stable (766.4.0) / k8s 1.0.1 / docker 
> running on Azure VMs)
>            Reporter: Michael Schonfeld
>            Assignee: Robert Joseph Evans
>            Priority: Critical
>             Fix For: 0.11.0
>
>         Attachments: Screenshot 2015-11-08 22.17.57.png, Screenshot 
> 2015-11-08 22.18.06.png
>
>
> We've been running Storm's snapshots on our production cluster for a little 
> while now (that back pressure support really helped us), and we've noticed a 
> sudden spike in system load when going from 
> commit@ba1250993d10ffc523c9f5464371fbeb406d216f to the current latest 
> commit@c12e28c829fcfabc0a3a775fb9714968b7e3e349. Both versions were running 
> the exact same topologies, and there was no significant change in workload. 
> Not exactly sure how to even begin to debug this, so we ended up just rolling 
> back. Thoughts?
> Stats screenshots attached



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

Reply via email to