Monitoring more than 1500-2000 servers

[rajsshah]

HI Team
We have to monitor around than 1500 - 2000 servers . Each server having 5-10 service checks.We wanted to have a suggestion on below topics for our nagios monitoring platform.
We are using Azure virtual machine , modgearman . ( 2 nagios machine [ 1 active at a time using drbd concept ] , seperate mariadb database , 1 modgearman worker on seperate machine )

My question :
- What is the hardware you recommend ( in terms of memory , CPU , machine type )
- NCPA or NRPE which one we should select.
- NRPE is not supported on windows ??
- Whether Modgearman supports NCPA ??
- whether we should have ACTIVE checks or PASSIVE checks . Which can cause more impact to nagios server performance.
- Are all NRPE plugins can be run via NCPA ??

[benjaminsmith]

Hello @rajsshah,

I've included some guidelines below for the recommended hardware requirements and our general guide for maximizing system performance.

Hardware Requirements
Maximizing Performance In Nagios XI

Your Questions:
1. Increasing the number of CPU cores and RAM beyond the guidelines will certainly help boost performance (e.g 16 RAM / 8 Cores ). Generally speaking you should have at least 50% more memory than needed by the system.
2. I would recommend using NCPA.
3. That's correct. NRPE cannot be installed on Windows.
4. You can use Mod_Gearman with NCPA. Mod-Gearman is a broker module that integrates with Nagios Core and helps offload some of the processing to other workers. However, it does make your installation more complex.
5. Passive checks require less resources from the Nagios Server. ( see: Configuring Passive Services with Nagios XI ). However, configuring active checks to check only as often as necessary will help decrease the load as well.
6. Yes. This is possible. You would need to compile the Nagios Plugins on the Linux system then put them in the NCPA plugins directory.

Let me know if need clarification on anything or have any other questions.

[rajsshah]

Hello @benjaminsmith

First of all thanks for the response . Much appreciated . While I am going through the document I have 1 query about your comment

"You would need to compile the Nagios Plugins on the Linux system then put them in the NCPA plugins directory."

what do you mean by compile exactly ?? If I have a nrpe plugin with *.pl file or *.sh file . I don;t have to compile any thing correct ??? I am a bit confused .

[benjaminsmith]

Hi @rajsshah,

Yes, that is correct. If you have a shell script or pearl plugin, you can just add this to the NCPA plugins directory. However, many of the Nagios Plugins in are written in C, may have other dependencies, and will need to be installed from source.

Nagios Plugins - Installing Nagios Plugins From Source

[rajsshah]

Hello @benjaminsmith
Thanks for info .I have some queries regarding hostgroup and servicegroup. For monitoring more than 1000 servers , I created different hostgroup which works fine . However for service group I am a bit confused on how to use it efficiently and how it can benefit me . I was hoping that there will be mapping b/w hostgroup and servicegroup as well . but I don't think so it is possible ..isn't it ???

Also consider this example : I have 100 servers in hostgroup A . I have created a filesystem check service ( using check_ncpa.py) and assigned it to hostgroup A. This filesystem check service is something like below

$USER1$/check_ncpa.py -H $HOSTADDRESS$ $ARG1$

ARG1 : -t xxxxxx -M 'disk/logical/$_HOSTMOUNT_POINT$/used_percent' -P 5666 --warning 80 --critical 95

_MOUNT_POINT variable is defined at every host level so that they can pass which mountpoint they want to monitor . This is working fine.

Now out of 100 server , I have 5 server where in apart from "/" fileystem I also want to monitor "/opt/xxx/" filesystem . Is there a possibility to achieve this with same service ?? I know I can achieve this by creating new service and assigning it to those 5 hosts , but that I will consider dirty . It could be that some servers want more than 1 or 2 mount points to monitor , then again I will have to create more services ?? is there any concept of loop in service ??

[benjaminsmith]

Hi @rajsshah,

Now out of 100 server , I have 5 server where in apart from "/" fileystem I also want to monitor "/opt/xxx/" filesystem . Is there a possibility to achieve this with same service ?? I know I can achieve this by creating new service and assigning it to those 5 hosts , but that I will consider dirty . It could be that some servers want more than 1 or 2 mount points to monitor , then again I will have to create more services ?? is there any concept of loop in service ??

That's not part of the standard Nagios XI functionality in the web interface. Aside from creating new service checks, you'd have to develop a custom solution/script.