Over the last couple of weeks I’ve had the chance to start playing around with omd. This is an opensource project combining nagios, check_mk, php4nagios and some other stuff that might be interesting (not tested those last ones).
I started using it after playing with check_mk (an earlier version). Had to install it by hand and found out that omd currently has packages for Linux distro’s like Suse, Debian and some others. Since a Debian packege was available I decided to give it a try. The installation was a breeze. Just needed to download the .deb file and it delivered all I needed. Since I am only using it for our own company I am not worried about certain versions the software needs. It is running on a separate monitoring lan so nobody besides the people that need to monitor is able to use it.
The product itself is able to configure several sites that will be able to monitor segments as seperate users. It will start seperate apache instances giving you the potential of giving certain customers an instance where they can only see their own machines. All of the sites can be configured to create al livestatus link where one instance can combine 1 to N sites that will be monitored by one single livestatus instance. Not sure if this is a secure setup but we plan to investigate this in the near future.
The interface. Omd is giving you a general command line interface that will install instances that are very simple to handle. It is able to start, stop, restart, copy, remove, backup, etc.. instances with single line commands. One downside, if you want to copy an instance you need to stop the instance you want to copy.
The command line interface has two users. The omd user runs as root, the instance user runs as <instance name>. General commands are run as root (or as omd if you know what you are doing). Site specific commands are run as the user <site instancename>
The frontend contains some webinterfaces that will allow you to run basic operations like adding hosts, deleting hosts, decide what a host is (linux, windows, network version, snmp version) within check_mk (one of the products in an instance). It can scan the services and let you add seperate services within the scan to tag and monitor. It can also restart the instance if you need to activate changes (that is a bit silly, not needed imho if nagios would allow a dynamic config).
Performance. Omd is able to setup instances using the temp filesystem and loading this into memory as far as the hardware allows this. Where I needed a vm with 3 cpu’s and 4 gigs of mem I am now able to run 30 to 40 percent more service checks on a machine with 2 cpu’s and 2 Gb memory. I suspect I can tune this even more but I have some legacy checks that might be consuming memory and cpu. The load for 3000 plus services is very low (0.10 average).
Checks are mostly based on snmp and bulkwalk. If you do know the python and nagios syntax it will allow you to use legacy checks, new snmp checks, etc.. I am using the legacy checks to monitor IIS and MS SQL performance counters. I also monitor AD but that is mostly done by checking ports. For most of the Unix/Linux machines the normal checks are sufficient. You will have base filesystem, cpu and memory checks but describing services in the config will also allow you to monitor minimum/maximum instances of unix processes, forked or not.
For those interested in trying it: