Skip to content

Instantly share code, notes, and snippets.

@pburkholder
Created February 27, 2014 16:16
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save pburkholder/9253297 to your computer and use it in GitHub Desktop.
Save pburkholder/9253297 to your computer and use it in GitHub Desktop.
Hi Sensuite:
I'd like each node to check our graphite server to determine if its CPU steal percent merited a warning. Of course, when I tried it, that generated a thundering herd when the sensu server published a check request. I was doing this because the check_cpu.rb plugin has steal/usage that are 10x what we get from collectd (or what I observe w/ top).
I can re-implement as a standalone check with some sort of randomization around the interval, but I prefer most checks regarding node health to come from the server.
From https://github.com/sensu/sensu/blob/master/lib/sensu/client.rb#L221 it looks like one could add a random splay to `scheduling_delay` without too much hackery, but that may be undesirable.
So this brings me to the following questions:
* Is there a current method to splay the client scheduling of checks?
* Would adding splay be an acceptable feature?
* Am I just wrongheaded about this?
* (and why does check_cpu.rb give such inflated results)?
-Peter B
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment