Nagios monitoring system

derfderfley · 17 Nov 2011 at 20:22

Anyone got any idea of hardware requirements for a large scale Nagios install? by large I'm talking around 1500 hosts and 6-7000 services.

AndrewP · 18 Nov 2011 at 10:38

We've got 1200 hosts and 1600 services on a Dell 2950, 2 quad-core 2.5 GHz Xeons (L5420), 4GB RAM and 15K SAS disks in RAID 1. It also runs PNP for about half the services and some of the service checks are pretty complex. The server is just about coping with CPU being the bottleneck.

AlexD · 18 Nov 2011 at 14:58

I use a 2 x opteron 256 box which is monitoring 1000 hosts with about 7000 services monitored and it copes fine. I do use check_mk addon for nagios which is main reason it performs so well.

bods · 20 Nov 2011 at 08:25

I run the same size environment on a VM - 4GB ram, 2vCPU. However if you want to include pnp then that will add significant workload to the server - probably need to look at doubling the RAM and CPU and make use of RAM disk.

Suggest you also look at the documentation out there on optimising Nagios for larger environments.

Stolly · 20 Nov 2011 at 09:34

Love Nagios. Its free and better than a lot of the monitoring systems you'd pay many thousands for.

derfderfley · 20 Nov 2011 at 18:54

I have a feeling disk performance is going to be key as it's running centreon as well. Probably going to back off the amount of data stored in the database to a minimum 30 day'ish retention, and increase the retention of the RRD files to 360 days. Hopefully that should stop the database growing too big and causing too much disk thrash.

I've had a look through the large environment documentation a couple of times, and the only trouble is it is all a little too theoretical for my liking. Hence my original question about anyone having real world experience, and thanks for the replies so far

Volitous · 21 Nov 2011 at 10:39

do you need HA? secondary pollers? clustered mysql etc - obviously your resilience is limited if you putting this on one server

derfderfley · 21 Nov 2011 at 19:27

Volitous said:
do you need HA? secondary pollers? clustered mysql etc - obviously your resilience is limited if you putting this on one server

Well that's the other issue, and budget will bite on that count. :eek:

shine · 21 Nov 2011 at 19:34

Volitous said:
do you need HA? secondary pollers? clustered mysql etc - obviously your resilience is limited if you putting this on one server

I worked around this by using an old server as the redundant host - it only held monitored the basics of all the critical apps. Better than nothing in a DR type situation and life saver when we did network isolation tests.

clade · 22 Nov 2011 at 08:34

how do you guys decide what rules to use to monitor services? I want to give Nagios a try soonish but lack of metrics on some services make me pull my hair out.

SMN · 22 Nov 2011 at 12:51

Trial and error.
You can get it to do pretty much whatever you want.