Virtual Nagios and high disk rate

[mguthrie]

Just an FYI, we'll be pushing out the new 2.3 release on Monday.

[TenTimes]

Thanks for the information.

I installed the update yesterday and it seems to me that the disk rate problems have lessened. We now get a small spike every 2 minutes instead of a huge spike every minute, so cron seems to have been improved (still haven't seen the disk rate go over a 1000 Kb/s). The apache service seems to be using a bit less RAM now and it doesn't grow over time as much as before the update.

There is a small increase in CPU usage, but the bennefits outweigh this problem. I still have to see how the new update works after a week of usage, and how the growth in resource usage will turn out.

I'll keep you updated.

[mguthrie]

My observations on performance were similar to yours. The spikes seem to flatten out a bit, while the overall disk activity still remains around the same level. Thanks for the updates!

[TenTimes]

Just one question (off topic)

Is a usage of 1 complete processor (of 2) much for a machine which checks 70 hosts and about 2500 services? The average usage is about 27,5% of 2 processors. (virtual nagios)

[mguthrie]

There's actually quite a few things that can affect that question. My short answer is yes, that's probably normal. The variables in that are ultimately:

- How many checks are being run every second? ( ( hosts+services / average check interval ) / 60)
- This also depends on the checks that you're running. Compiled plugins take less CPU, while SNMP, ESX checks, and other plugins written in interpreted languages can take more CPU.
- Whether or not mysql is on the same server, or if it's offloaded to a second machine.

[TenTimes]

The sql database is on the same server and the checks are every 5 minutes.
The only plug-in we use is the ESX perl SDK and the NSClient++ on the servers (about 40 hosts)

[scottwilkerson]

This could be about right, however if your CPU usage is 27.5%, that's only about 1/2 of one processor, not 1 complete processor...

[mguthrie]

The only plug-in we use is the ESX perl SDK

That is actually a pretty big CPU user, so I'd say your system is running pretty nicely ; )

[TenTimes]

This could be about right, however if your CPU usage is 27.5%, that's only about 1/2 of one processor, not 1 complete processor...

This is the average CPU usage. It spikes every 1-2 minutes to about 50% then drops to around 20-22,5%.

We just stopped monitoring every port of our switches (not needed) and now we have about 800 checks and we now have a CPU usage of about 12.5-15% which is good.
Now I just have to solve the problem with the mysql database (in http://support.nagios.com/forum/viewtop ... 3&start=10).

Thanks for the help and you can close the tread now.