Wednesday, 18 April 2007

Nagios - The Final Word

Having posted about installing Nagios on CentOS I have finally had a few comments on BeerBytes (Hooray). I have also featured on 2 little news posts on other Linux sites as well click here for the latest and check out the blogroll right for a proper link. Given the obvious appetite for this subject (no not me - network monitoring) I suppose its only fair that I should continue dispensing nuggets of information given that its the only subject which has raised a flicker of interest.

Having got the basics set up I spent some time last week getting the network diagram straight, as we have quite a busy network with 60 nodes I wanted to monitor, the automatic function did not do it justice. Moving on from this I fell foul of having installed the incorrect plugin archive, for some reason check_ping worked fine but I wanted to start monitoring DNS, MySQL and HTTP on a couple of servers and these plugins would not run. In order to test your plugins you can simply move to the plugins directory and type ./pluginname, for example ./check_ping --help will tell you what command line parameters the ping plugin requires then try it again with these supplied. To continue the example ./check_ping -H www.yahoo.com -w 1000,10% -c 1000,10%
returns PING OK - Packet loss = 0%, RTA = 84.63 ms. The commands are already set up for the standard plugins in commands.cfg if your install was ok.


So to sort this problem out I ran off to try my old mate DAG's archive :o) I found the appropriate rpm and bingo, everything works well. If your are relativly new to Linux as I am and you are struggling to connect to the repository on yum or using rpm there is a quick and dirty work around. Simply find the link to the rpm in your web browser, copy the link, get back to your shell and type wget and paste the link, this will download it to your machine. Next type rpm -i and the name of the downloaded file and this will install everything, apologies if that was embarrisingly basic but yum can be a bit hit and miss for me.

The documentation for installing Nagios is quite good but the documentation for actually using some of the many and various standard plugins is really quite sketchy so look outside the Nagios.org site for this information. The Nagios plugins page on sourceforge is your starting point for this but its not obvious.

In order to start monitoring services a little editing of the services.cfg is required followed by creating a couple of new hostgroups for similar machines. For example I wanted to monitor Mysql on 2 machines so I declared the service in services.cfg as follows


define service{
use generic-service
name mysql-service
is_volatile 0
check_period 24x7
max_check_attempts 5
normal_check_interval 1
retry_check_interval 1
notification_interval 20
notification_period 24x7
notification_options n
check_command check-mysql-alive
service_description MYSQL
contact_groups nerds
hostgroup_name mysql_servers
}


This will check the hostgroup mysql_servers every minute, 24 hours a day and email the nerd group if there is a problem for 5 successive queries. It presumes a standard install which predefines the 24x7 time period and the check-mysql-alive command in the realvent files.

Now that I have these extra services being monitored there is some really quite useful informtion being generated, for example the number of connections connected to the MySQL servers and the response times for the HTTP servers. Another area which is being fine tuned via the timegroups.cfg file is when I want to be alerted about certain things, for example I quite like getting an email if a router goes down overnight but as some equipment is turned off overnight I dont particularly need to know. So in short just getting Nagios installed is the tip of the iceberg, the more you think about things the more instances where good network monitoring is useful become apparent. The good news is that this fine tuning is very quick and easy once you have the thing up and running.

One final point on Nagios before everyone gets bored is that on windows you can actually use active desktop to embed your live network map into your desktop making sure you never miss a trick :o) Simply go to desktop properties->customise and paste your Nagios address followed by /cgi-bin/statusmap.cgi?host=all into a new web address to embed into the desktop.

Some other things happening in our little team include a spontaneous upgrade to Adobe CS3, I haven't even got it installed yet so you will all have to wait for some views and opinion but one thing to remember is that you need bags of hard disk space, the download is about 1.5 GB and then it unpacks to nearer 2 GB then needs 5.6GB for the programs, so unless you want to be cleaning up after every stage you need about 10GB! Also the knowledge base is filling nicely and the more I use the product the more I like it, just one little point is that you have to keep going to different URL's to do different things, it does not check your security and give you all the options you are entitled to.

2 comments:

Anonymous said...

I gotta ask - you've got one great looking map compared to the rest of the people out there. What did you do to relocate your icons so nicely? Ours are in that hapless circle with layers that just don't really work for us.

Thanks

nk

MikeW said...

Once I had the proxy settings right, I had no issues with Yum.

1) in /etc/yum/conf, add a line 'proxy=http://abc.co.uk:8080'

2) for other apps, make sure env var 'http_proxy' is set to your proxy's URL.

To use Dag's stuff, you need to install the RPMForge settings.

I had the proxy as well, so my command ended up like:
sudo rpm -httpproxy proxy.abc.co.uk -httpport 8080 -Uhv http://apt.sw.be/redhat/el5/en/i386/rpmforge/RPMS/rpmforge-release-0.3.6-1.el5.rf.i386.rpm

Then you can just go:
(sudo) yum install nagios-2.9

Lots of useful day-to-day notes on sysadmin - hope we get some 2008 entries at some point ?!

A view from the rack is the personal blog of an IT manager who works for a pub company - hence beer