OT: A Lightweight Monitor Software

Home » CentOS » OT: A Lightweight Monitor Software
CentOS 16 Comments

Hi all,

I have a test lab installed over an ESXi 5.1 host that contains 5
CentOS vm guests. I would like to monitor CPU, RAM and Network for these vm and ESXi host. Zabbix and Nagios are too to accomplish this task. Does anyone know any lightweight openosurce soft to do this??


16 thoughts on - OT: A Lightweight Monitor Software

  • well,without any personal experience, what I heard about Munin was that it was quite easy to setup, and did a lot for you, but it was not ‘light weight’ by any stretch.

  • I use Munin.

    If you intend to monitor many hosts, that will lead to tons of charts to draw periodically. This is very I/O consuming, and you have to make sure you have enough for that, or spend some time to reconfigure the nodes to reduce monitored resources.

    Munin is easy to setup because it has configuration suggestions: it might monitor ressources that are useless for you.

  • You can also use munin charts in “on-demand” cgi mode where the charts are being drawn only when called for from the pregenerated HTML (which can also be on-demand). This also enables chart zooming. You need to configure your webserver to run few perl scripts via cgi.


    Otherwise munin-node is just a set of perl scripts and other than the dependency list, really is lightweight.

  • A relevant question here is whether you want monitoring in terms of alerts or trend graphing?
    This will help you narrow your software choices to what you need.

    Some monitoring software such as Zenoss does both (alerting and trend graphing) as do others. My current monitoring is a combination of Nagios
    (alerts) and Cacti (trend graphing).

    I’d be interested in other monitoring software you find and your impression of whatever you choose to go with.

  • Personally, I like OpenNMS but it is probably even bigger. But RAM
    is cheap these days and it is nice to start with something that will scale up if you need it later.

  • In my experience (and, I think, logically), lightweight packages are likely to require more labor to get the results that you want except for the rare occasion when the author’s goals and yours coincide.

    Nagios is actually quite light weight. I don’t like it much, but I use it extensively. Getting things like simple graphs out of it is rather difficult exactly because it is a light weight package.

  • you already have some great answers, but I want to throw in another susggestio : ganglia. Its provided in rpm form at EPEL ( including ganglia-web ), and if you have a trusted network, and multicast is considered ok – you wont need to do any config beyond install, and make 2 config changes in the main /etc/ganglia/gmond.conf file on each of the nodes