Re: Nagios : need some advice
* nicolas.kosmalski@alemtid.asso.fr <nicolas.kosmalski@alemtid.asso.fr> wrote:
> >> - Do i need to use another monitoring software?
> >I depends on what you want to monitor and what your boss wants to see.
> The essential things to monitor is switch, gateway ......
Bosses always appreciate - besides the uptime of your systems ;) - nifty
graphs visualizing internet link usage or the like. I found cacti
(www.cacti.net) to do a very good job at that.
Most of the time you don't know whats going on on your systems or your
network in terms of utilization. Cacti brings total transparancy to that
and supports you in long term capacity planning and retrospective
troubleshooting.
It does not do alerting (like nagios), but does a perfect job at
graphing measuring data (cpu load, memory / harddisk usage, link usage,
...). So it's a complement to nagios i consider worth using.
> >Every 10 seconds can put a lot of load on your network connections
> >especially if you monitor a lot of services and servers. I would
> >certainly not do that for most of the servers/services.
> So i will set a fast monitoring only for things like cpu load.
Do you think it is really necessary to monitor cpu load every 10
seconds? I don't. We check it at about every 5 minutes. Rather use
tresholds to do a more frequent checking when you get in the range of a
possible overusage. So normally check every 5 minutes but when you get
to a repeated 5-min load of 1 (for a 1 cpu system), do a more frequent
checking.
It is a lot about figuring out whats the best procedure for you specific
environment, but i definitly would not trigger a check every 10 seconds.
- sebastian
Reply to: