[Nagios-users] Installing Nagios Server on a Virtual Machine

Discussion:

Juki

2009-11-12 11:42:47 UTC

Hello people,

I would like to know if it is advisable (or best practice) to install and
run a Nagios monitoring server on a virtual machine (in this case, with
OpenSuSE as the OS) with
the intention of monitoring physical hardware client machines on the same
LAN.

If so, what known issues should I look out for in this case?

Thanks,
Juki

Christian Schneemann

2009-11-12 12:03:05 UTC

Permalink

Hi,

Post by Juki
Hello people,
I would like to know if it is advisable (or best practice) to install and
run a Nagios monitoring server on a virtual machine (in this case, with
OpenSuSE as the OS) with
the intention of monitoring physical hardware client machines on the same
LAN.

we have our Nagios testsystem and some distributed Nagios' running in Xen
guests. We are seeing no problems with doing that.

Post by Juki
If so, what known issues should I look out for in this case?
Thanks,
Juki

DE/HAM Hoppe, Leif

2009-11-12 12:59:32 UTC

Permalink

Hi Juki,

No problems here, either.
OpenSuSe on Vmware ESX.

regards from Hamburg

cheers
Leif

-----Original Message-----
From: Christian Schneemann [mailto:***@suse.de]
Sent: Donnerstag, 12. November 2009 13:03
To: nagios-***@lists.sourceforge.net
Subject: Re: [Nagios-users] Installing Nagios Server on a Virtual Machine

Hi,

we have our Nagios testsystem and some distributed Nagios' running in Xen
guests. We are seeing no problems with doing that.

Post by Juki
If so, what known issues should I look out for in this case?
Thanks,
Juki

Frost, Mark {PBG}

2009-11-12 17:36:38 UTC

Permalink

This was a year or two ago, but we found that when we ran Nagios in this way it worked in general, but because of the sort of variable size of a second on VMware, the latencies were kind of screwed up. This was clearly evidenced when we looked at the performance statistics. Nagios indicated that a lot of checks ran earlier or later than it had expected them to.

I don't know if somehow that's gone away or not, but it was a big issue for us and not within the realm of things we were able to tolerate so we want back to physical servers.

One of the arguments I know I've seen before on this list is the idea that you're doing your critical system monitoring inside an abstracted layer (the VM) which might alter your view of the world or fail to work should there be an issue with the ESX server. But all possible acceptable depending on your site's needs.

Mark

Post by DE/HAM Hoppe, Leif
-----Original Message-----
Sent: Thursday, November 12, 2009 8:00 AM
Subject: Re: [Nagios-users] Installing Nagios Server on a Virtual Machine
Hi Juki,
No problems here, either.
OpenSuSe on Vmware ESX.
regards from Hamburg
cheers
Leif
-----Original Message-----
Sent: Donnerstag, 12. November 2009 13:03
Subject: Re: [Nagios-users] Installing Nagios Server on a Virtual Machine
Hi,

Post by Juki
Hello people,
I would like to know if it is advisable (or best practice) to install

and

Post by Juki
run a Nagios monitoring server on a virtual machine (in this case,

with

Post by Juki
OpenSuSE as the OS) with
the intention of monitoring physical hardware client machines on the

same

Post by Juki
LAN.

we have our Nagios testsystem and some distributed Nagios' running in Xen
guests. We are seeing no problems with doing that.

Post by Juki
If so, what known issues should I look out for in this case?
Thanks,
Juki

Greetings,
Christian
--
Christian Schneemann
Operations & Services
-------------------------------------
SUSE LINUX Products GmbH,
Maxfeldstr. 5, D - 90409 Nürnberg
Phone: +49 (0)911 - 740 53 0
-------------------------------------
SUSE LINUX Products GmbH, GF: Markus Rex
HRB 16746 (AG Nürnberg)
------------------------------------------------------------------------
------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-
Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july
_______________________________________________
Nagios-users mailing list
https://lists.sourceforge.net/lists/listinfo/nagios-users
::: Please include Nagios version, plugin version (-v) and OS when reporting any issue.
::: Messages without supporting info will risk being sent to /dev/null
------------------------------------------------------------------------
------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-
Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july
_______________________________________________
Nagios-users mailing list
https://lists.sourceforge.net/lists/listinfo/nagios-users
::: Please include Nagios version, plugin version (-v) and OS when reporting any issue.
::: Messages without supporting info will risk being sent to /dev/null

gmartin

2009-11-13 01:18:37 UTC

Permalink

This definitely a question of scale. I suspect a virtual solution could
support a couple thousand services or even more in a distributed
environment. For a smaller environment it would be a no-brainer.

We virtualize everything by default so the next nagios server will go
virtual in our MS Hyper-V pool.

\\Greg

On Thu, Nov 12, 2009 at 12:36 PM, Frost, Mark {PBG}

Post by Frost, Mark {PBG}
This was a year or two ago, but we found that when we ran Nagios in this
way it worked in general, but because of the sort of variable size of a
second on VMware, the latencies were kind of screwed up. This was clearly
evidenced when we looked at the performance statistics. Nagios indicated
that a lot of checks ran earlier or later than it had expected them to.
I don't know if somehow that's gone away or not, but it was a big issue for
us and not within the realm of things we were able to tolerate so we want
back to physical servers.
One of the arguments I know I've seen before on this list is the idea that
you're doing your critical system monitoring inside an abstracted layer (the
VM) which might alter your view of the world or fail to work should there be
an issue with the ESX server. But all possible acceptable depending on your
site's needs.
Mark

Post by Juki
Hello people,
I would like to know if it is advisable (or best practice) to install

and

Post by Juki
run a Nagios monitoring server on a virtual machine (in this case,

with

Post by Juki
OpenSuSE as the OS) with
the intention of monitoring physical hardware client machines on the

same

Post by Juki
LAN.

we have our Nagios testsystem and some distributed Nagios' running in Xen
guests. We are seeing no problems with doing that.

Post by Juki
If so, what known issues should I look out for in this case?
Thanks,
Juki

Greetings,
Christian
--
Christian Schneemann
Operations & Services

\\Greg

T***@csiro.au

2009-11-13 01:55:53 UTC

Permalink

Hi

We have an esx pool of about 120 VMs, and about 4-5 service checks per vm (disk/stock web/ssh/ldap/ssl certificate expiry/tomcat web apps/etc etc). The nagios monitor (also a vm) hits them all.

Only ‘unique’ issue we have since the vms are in a really aggressive load balancer, outages in networks with effect different VMS as they have vmotioned somewhere else in the cluster. We haven’t won that battle – and are happy to just live with it for now.

As for scaling – its a debian box with 256mb of memory also running our LDAP slave server, and looks at 600 different things. And we run nagiosgrapher to disk.
I think thats a win for nagios ☺

Terry

From: gmartin [mailto:***@gmartin.org]
Sent: Friday, 13 November 2009 9:19 AM
To: nagios-***@lists.sourceforge.net
Subject: Re: [Nagios-users] Installing Nagios Server on a Virtual Machine

This definitely a question of scale. I suspect a virtual solution could support a couple thousand services or even more in a distributed environment. For a smaller environment it would be a no-brainer.

We virtualize everything by default so the next nagios server will go virtual in our MS Hyper-V pool.

\\Greg
On Thu, Nov 12, 2009 at 12:36 PM, Frost, Mark {PBG} <***@pepsi.com> wrote:
This was a year or two ago, but we found that when we ran Nagios in this way it worked in general, but because of the sort of variable size of a second on VMware, the latencies were kind of screwed up. This was clearly evidenced when we looked at the performance statistics. Nagios indicated that a lot of checks ran earlier or later than it had expected them to.

I don't know if somehow that's gone away or not, but it was a big issue for us and not within the realm of things we were able to tolerate so we want back to physical servers.

One of the arguments I know I've seen before on this list is the idea that you're doing your critical system monitoring inside an abstracted layer (the VM) which might alter your view of the world or fail to work should there be an issue with the ESX server. But all possible acceptable depending on your site's needs.

Mark

Post by Juki
Hello people,
I would like to know if it is advisable (or best practice) to install

and

Post by Juki
run a Nagios monitoring server on a virtual machine (in this case,

with

Post by Juki
OpenSuSE as the OS) with
the intention of monitoring physical hardware client machines on the

same

Post by Juki
LAN.

we have our Nagios testsystem and some distributed Nagios' running in Xen
guests. We are seeing no problems with doing that.

Post by Juki
If so, what known issues should I look out for in this case?
Thanks,
Juki

Greetings,
Christian
--
Christian Schneemann
Operations & Service

Steve Shipway

2009-11-13 01:31:56 UTC

Permalink

I would be very very wary of running Nagios on a VM (we use VMware here). The reason for this is Clock Skew.

Clock Skew causes the virtual clock on the guest OS to lag behind then skip forward depending on the loading and sleep times of the guest. Note that this will not affect 'Para-virtualised' guests as they share the hardware clock, but these are only possible in some Xen guests at the moment AFAIK and are not common. VMWare can't do them.

On a lightly loaded physical machine your clock skew will be negligible but as load goes up you can get the guest clock lagging as much as 10sec or even more. This can screw up latencies, scheduling, rate calculations (such as CPU use and net use) and so on. In addition any monitoring of virtualised resource (CPU, Memory) will be completely wrong unless you obtain the values from a source which is aware of the virtualiasation (eg VMWare tools API or VirtualCentre API for vmware)

Clock skew and virtualised resource monitoring has caused too many problems in our tests and we now only use physical servers for Nagios (and MRTG).

I have a Nagios plugin check_vmware at www.steveshipway.org/forum<http://www.steveshipway.org/forum> for monitoring VMware virtualised resources via the API to get meaningful values - previous used check_esx3 but this has been superceeded by the use of the VC API in check_vmware

Steve

________________________________
From: Juki [***@gmail.com]
Sent: Friday, 13 November 2009 12:42 a.m.
To: Nagios Users Mail-list
Subject: [Nagios-users] Installing Nagios Server on a Virtual Machine

Hello people,

I would like to know if it is advisable (or best practice) to install and run a Nagios monitoring server on a virtual machine (in this case, with OpenSuSE as the OS) with
the intention of monitoring physical hardware client machines on the same LAN.

If so, what known issues should I look out for in this case?

Thanks,
Juki