Nagios Support Forum

I/O Wait 0.10%

What do you mean by:
"You may want to make sure the "check_ifoperstatus" service checks are configured correctly."

Everything was configured by default, using the wizards more or less.

Like I wrote on a previous post I have 80 hosts and 5000 checks.
Some of my hosts and interfaces are down and some are on slow links and I get lots of flapping. I need to fine tune this (but for now I can live with that).

My problem is that I had info on the graphs until 15 days ago, on all the monitored services and interfaces, and out of a sudden I just lost that info...

rcordeiro wrote: What do you mean by:
"You may want to make sure the "check_ifoperstatus" service checks are configured correctly."

Judging by your ps output, this check is using a high amount of cpu compared to other checks.

Like I wrote on a previous post I have 80 hosts and 5000 checks.
Some of my hosts and interfaces are down and some are on slow links and I get lots of flapping. I need to fine tune this (but for now I can live with that).

That is near upper limit as far as our hardware specs document is concerned. But nothing in top or ps eludes to why load is so high.

My problem is that I had info on the graphs until 15 days ago, on all the monitored services and interfaces, and out of a sudden I just lost that info...

That is most likely when the server's load became too much for npcd and perfdata stopped processing.

OK,

Any light you can shed on me about what I can do to get graphs data again?
Right now I can change almost eveything on the system and data on the graphs is a must.

Regards and thanks for all your help.

Either we hunt down the load issue (using top and ps liberally), or you increase the hardware. If load gets too high, performance data will stop processing. We could increase the load threshold, but it already is set VERY high. Did you change or add any checks 15 days ago?

We have been adding hosts and services but not a bunch of them in a particular day, and most of them have been added before that.

Can the devices that give timeout, or the services down have an impact on this?

I can disable some checks on interfaces I now it will be down for some time, just to check if the systems starts to give me some data on the graphs.
You think it's just a matter of performance?

The graphs not showing up is most likely performance. It is hard to test if it is something else because the load is too high. The question is "why is the load so high?"

Can you post a screenshot (or copy the data) from the "System Status" page under "Admin" in XI?

Hi,

Here are two shots from the System Status, one from yesterday and another one from today:

It would be helpful if we could also see Admin -> Monitoring Engine Status

As a refresher how many CPU cores does this server have?

thanks

4 Cores

One thing to note, my colleague is updating all the services and hosts with specific information on the description and GPS coordinates. This also creates a lot of updates and eventually lots of CPU. Never the less the processes hat use the most CPU time are the mrtg interfaces check.

Ok, this may clear some things up for for why the load is so high at times. Also, looking at the Server Statistics, leads me to believe that the system may have multiple threads per core if this is a 4 core system, as it is still showing green with the load of 20...

In the end, have your graphs started showing data?