[THIN] Re: Monitorig Config

  • From: Magnus Hjorleifsson <magnus@xxxxxxxx>
  • To: "thin@xxxxxxxxxxxxx" <thin@xxxxxxxxxxxxx>
  • Date: Tue, 16 Nov 2010 18:24:44 -0500

I sent mine to 7 to 10k per processor / core in the server if physical. For 
virtual I set it to 5 to 7k per vCPU. That seems to work for us pretty well for 
80% of the apps we have running

Sent from my iPhone

On Nov 16, 2010, at 14:34, Berny Stapleton <berny@xxxxxxxxxxxxxxxxx> wrote:

> I'm getting a LOT more than that because of a single process that's doing IO 
> and not writing it to shared memory. Also, you will see a lot more on 
> multi-core systems where applications give up their processor time to request 
> data from memory on another CPU.
> As Joe said, find out what's a safe working load on your servers, and then go 
> above that to detect problems or changes to the environment.
> On Tue, Nov 16, 2010 at 5:53 PM, Greg Reese <gareese@xxxxxxxxx> wrote:
> Just curious what y'all use for monitoring and alerts settings.
> I am getting a lot of alarms on Context Switches/sec but everything seems 
> fine on my servers.  I may just have it set wrong.
> I have yellow set at 12000 and red at 14000.
> No clue if that is appropriate or not.

Other related posts: