Check_local_load arguments

Support forum for Nagios Core, Nagios Plugins, NCPA, NRPE, NSCA, NDOUtils and more. Engage with the community of users including those using the open source solutions.
Locked
johjd94651
Posts: 8
Joined: Wed Apr 23, 2014 1:02 pm

Check_local_load arguments

Post by johjd94651 »

I'm struggling with the check_local_load arguments. By default nagios generates a "check_local_load -w 5.0,4.0,3.0 -c 10.0,6.0,4.0" command. As I understand this, the numbers indicate a relative processor load equivalent to 1.0 per processor or warning at a 5,4, and 3 CPU load equivalents over 1,5, and 15 minute averages. Also return a critical warning at 10,6, and 4 CPU load equivelents over the same 1,5, and 15 minute averages.

Weird thing is that these thresholds appear relatively standard over the documentation I've surveyed on google, -w 5.0,4.0,3.0 -c 10.0,6.0,4.0. Why would the default setting presume 10 cpu's? In my NAGIOSXI vm there are only 2 cpu's assigned to the server, yet it is returning warnings that apparently show a higher cpu load than there are processors assigned to the server.

I've read
http://support.nagios.com/forum/viewtop ... 66&p=49342
but it does not seem to hold my answer.

What am I misunderstanding here?
tmcdonald
Posts: 9117
Joined: Mon Sep 23, 2013 8:40 am

Re: Check_local_load arguments

Post by tmcdonald »

That plugin and the arguments make no assumptions about the number of CPUs. A load of 10 on a system with 2 CPUs is effectively the same as a load of 1.0 on a system with a single CPU. But a load of 10.0 on a system with 10 CPUs is equivalent to a load of 1.0 on a system with a single CPU, and equivalent to a load of 0.5 on a system with 20 CPUs.

To quote abrist from that thread:
abrist wrote: A load of "1.0" is one core and 100% percent utilization. Anything over implies that there was some wait involved. But if you had 4 cores, a load of "1.0" would imply that on average, all 4 cores are working at 25% of capacity with no wait. At load "4.0", the quad-core system will be at 100% utilization and anything over will imply that there was some wait.
Former Nagios employee
johjd94651
Posts: 8
Joined: Wed Apr 23, 2014 1:02 pm

Re: Check_local_load arguments

Post by johjd94651 »

Thanks for the response tmcdonald. However, that does not clear it up for me. Are you saying that 10 = 1.0 in terms of syntax? I don't understand how a load of 10 for 2 cpu's = a load of 1 for 1 cpu = a load of 10 for 10 cpu's = a load of .5 over 20 cpu's? I would have thought for example that a load of 10 on 2 cpu's would = 500% utilization.
tmcdonald
Posts: 9117
Joined: Mon Sep 23, 2013 8:40 am

Re: Check_local_load arguments

Post by tmcdonald »

I'm really not sure why I typed that... Let me fix that (corrections emphasized):
tmcdonald wrote:A load of 2.0 on a system with 2 CPUs is effectively the same as a load of 1.0 on a system with a single CPU. And a load of 10.0 on a system with 10 CPUs is equivalent to a load of 1.0 on a system with a single CPU, and equivalent to a load of 0.5 on a system with 20 CPUs. Essentially, (Load shown / # of CPUs) = effective load.
Sorry for the confusion.
Former Nagios employee
johjd94651
Posts: 8
Joined: Wed Apr 23, 2014 1:02 pm

Re: Check_local_load arguments

Post by johjd94651 »

Ok, that's what I thought, so on a 2 cpu system, a 1.8 would be a 90% load. Which is why I could not figure out why I was getting warnings on the tactical overview for a 2 cpu VM set with the following config: check_local_load -w 5.0,4.0,3.0 -c 10.0,6.0,4.0 As far as i can tell from the system performance stats on the vm itself, .17 is the highest the vm has actually gone.
tmcdonald
Posts: 9117
Joined: Mon Sep 23, 2013 8:40 am

Re: Check_local_load arguments

Post by tmcdonald »

Glad I could clear that up for you, and sorry again for the miscommunication.

I'll be closing this thread now, but feel free to open another if you need anything in the future!
Former Nagios employee
Locked