Recommended check_interval

[amitgupta19]

I am using the Nagios Core 4 installed on CentOS 7.

With following configuration:
1) MYSQL DB
2) 500 Hosts
3) 5000 Service Definition

Server is the Physical server and having the following configuration:
RAM: 32 GB
CPU: 2 Cores

Please suggest what should be the check_interval for the Hosts and services.

[eloyd]

Why is this machine running Core 4, but your machine in https://support.nagios.com/forum/viewto ... =7&t=48454 is running Core 3? At any rate, just as the other one, you're not asking a specific question. Checking to see if ICMP packets are working can be done every second without much harm. Running a complex SQL query to check linked databases and other internal structures while performing outside joins may take a huge amount of resources and only be done once every 15 minutes.

Different service checks can have different intervals, and it is all dependent upon what those checks are doing. Also, the specs on your database machine are likely to be more important here than the specs on your Nagios box. After all, it will be the database machine responding to the queries (and doing the work). Nagios will just be scheduling the tasks.

[amitgupta19]

Currently i have all the Monitoring done by the Nagios Core 3.

Now i am in process of setting up of new server that i am talking about here.
Slowly i will move all the monitoring of the Nagios Core 3 to the new server discussed here.

Do you mean if i decrease the check_interval to the 2 or 1 minutes for all the Hosts and services, will not make any impact on the nagios performance?

I am trying to find out any recommendations for the 500 Hosts and 5000 services for the check_interval.

[eloyd]

I do not mean that you can set all of the check_intervals to 1 or 2 minutes with no impact. I simply mean that it will depend on what those service checks are doing. If they're just checking ICMP packet loss, then yes, you could set it to one minute. If it's doing very complex SQL work, then 1 minute will likely kill your Nagios box and SQL server.

We would need more information on what checks you are tying to perform.

[amitgupta19]

It is mix of all
1) Drive Space for Windows and Linux Servers
2) UP Time
3) Memory Usage
4) Service State
5) NFS Mount
6) CPU Usage on Linux and Windows Server


Above are the standard parameters that we are using for all the servers.

[eloyd]

So none of these are doing SQL probes within a database or other complex checks, so I'd start with 5 minute intervals for the transient things (memory, CPU, load), 15 minute for other things like drive space and uptime, and see what happens after a few hours.

[amitgupta19]

Thanks for the clarification.

Can you please help me understanding "SQL probes within a database or other complex checks" .

What type of the monitoring it is?

[eloyd]

I was using a made up example of having to monitor an SQL server with a complex SQL query to check internal database connections and other things. If you had a need to do that, you would know what I was talking about!

[tmcdonald]

To add some explanation regarding tuning check intervals:

The question should not be "How fast can/should I make it?" but rather "How fast does it need to check?". A ping check makes sense to run more often than a check for OS updates - a ping is a more lightweight operation, can be relied on as a basic check of availability, and OS updates might only be pushed every 15 minutes at their fastest (on average) so checking for them every minute is wasted cycles.