Nagios Support Forum

Hello, I'm running v5.6.8 on a 64-bit CentOS 7 VM.

I have about 900 servers defined. 17,800 probes.

My server has 12 GB of memory, 1 GB swap space.

630 MB of swap space is in use, it seems to be slowly increasing.

Is 1 GB enough swap space for a 12 GB server?

Is there any known memory leak with 5.6.8 ?

Earl

Is 1 GB enough swap space for a 12 GB server?

Probably not - see the recommendations, listed in the Red Hat documentation here:

https://access.redhat.com/documentation ... -swapspace

Having said that, the question is why the system is swapping on the first place. Are you allowing memory sharing across virtual machines? This is a VM, so I wonder if the resources are shared, which can cause the issue.

Also, you have a quite large install. What type of checks are you running - mostly active or passive, SNMP, vmware checks, etc.?

Have you done any "fine-tuning" of you Nagios XI server to increase the performance? We generally recommend installing Nagios XI on a physical server when having a very large environment. Please, read our document on hardware requirements here:

https://assets.nagios.com/downloads/nag ... ements.pdf

Is there any known memory leak with 5.6.8 ?

Not that I know of. There was a memory leak in the old versions of Nagios Core (4.4.3 to be exact), which was fixed in 4.4.4.

https://github.com/NagiosEnterprises/na ... /Changelog

However, the version that you are using should have Nagios Core 4.4.5 as the underlying engine.

Hello,

We are running a lot of NRPE checks.

I don't know about the VM memory issues, I can ask around.

The load is often close to 2.0.

We use a RAM disk.

We will probably not have much increase in load. Most new servers eventually replace an equal number of old servers.

My Nagios Core (through the "View in Nagios Core" option) displays the version 4.4.5.

Hello,

We do have a few hundred probes that keep timing out (some 30 second, some 60 second) without much priority in resolving.

Can that be a performance problem?

Earl

Bitflogger wrote:Hello,

We do have a few hundred probes that keep timing out (some 30 second, some 60 second) without much priority in resolving.

Can that be a performance problem?

Earl

It's really not too likely that just a couple hundred would make that much difference. There is an impact but mainly only when they initially fail.

Hello,

I restarted a non-Nagios XI process, and my swap space went down considerably.

Please lock the case.

Thanks!

Earl