Sergio Belkin wrote:

2008/5/12 Les Mikesell lesmikesell@gmail.com:

...

Sergio Belkin wrote:

...

Hi, I'm looking for a monitoring system that support snmp v3. I want to monitorize linux servers and network switches. Currently, I am trying to use zabbix, but sadly, it lack at present features that I need. For example, I want to get reporting screens with data and graphs from network switches, I'd like to configure one only port of a given switch and that is used as templates for the rest of switch ports and the rest of the switches.

I'd like to use some open source software that meet that features, and I want to avoid Nagios :)

Could you recommend me someone?

OpenNMS will do most of this automatically if the snmp setup is the same on all the devices. Http://www.opennms.org. Installing from the yum repo that includes Sun java is the easiest approach.

Thanks Les, OpenNMS sounds interesting, in order to monitor a network switch, should I write XML files by hand?

There are a few things you still have to edit by hand, but development is very active and most of the configuration has been moved into the web interface. If your switches are common brands, the MIBs will already be included, and if you set the snmp collector to store values for 'all' interfaces it will build graphs for them automatically - or you can let it detect nodes, then follow the admin link for the node and pick the interfaces to collect. If you don't like the default graphs or want to add more, you might need to modify the xml files that describe them. but to start out, just set the snmp defaults and a discovery range and see what it does.

2008/5/13 Kai Schaetzl maillists@conactive.com:

Sergio Belkin wrote on Mon, 12 May 2008 23:07:20 -0300:

...

[CentOS] Somewhat OT:

even then please write a senseful subject next time!

Kai

-- Kai Schätzl, Berlin, Germany Get your web at Conactive Internet Services: http://www.conactive.com

Yes, you're roght Kai, I don't know how I could write such a stupid subject, but it was too late yersterday, and I was writing with a little part of my brain working :)

Even so, thanks for your comments, I'd like more experiences about monitoring systems. Again of topic, I want to avoid Nagios because it looks like over complex but if someone has an actual experience demostrating the opposite, I'd be glad to hear.

Thanks in advance

2008/5/13 jleaver+centos@reachone.com:

Sergio Belkin wrote:

...

Even so, thanks for your comments, I'd like more experiences about monitoring systems. Again of topic, I want to avoid Nagios because it looks like over complex but if someone has an actual experience demostrating the opposite, I'd be glad to hear.

Thanks in advance

We've used Nagios very successfully. We have hundreds of hosts and well over a thousand checks, so I'm guessing that we're probably a medium-ish installation. The use of templating makes adding hosts and services quick and painless. We've evaluated some of the other options already mentioned here: zabbix, opennms, zenoss, even mon, and big-brother and friends, and have always decided that nagios is the best product for our needs, as far as system monitoring goes. The initial learning curve is about medium compared to some, and once you've gotten over that hump, there just don't seem to be others. I've recommended Nagios to a few less-than-seasoned sysadmins who were able to take the templating concept and run with it. We have also setup cacti for the snmp statistics keeping. Nagios does have performance data capabilities now, they feel sort of tacked on to me. The folks over at http://www.centreon.com/ are working on an integrated user interface that includes statistics keeping using Nagios as the monitoring engine which looks as though there may be some promise, if I was starting over I'd definitely evaluate that.

I hope this is of some help in your review process.

Sincerely,

Jacob Leaver Sr. Systems Administrator ReachONE Internet _______________________________________________

OK, you won :) I'm going to test nagios. I am using centos 5.1 x86_64. Do I lose much if I use rpm from rpmforge (version 2.9)?

Sergio Belkin wrote:

2008/5/13 jleaver+centos@reachone.com:

OK, you won :) I'm going to test nagios. I am using centos 5.1 x86_64. Do I lose much if I use rpm from rpmforge (version 2.9)?

We're running version 2.11 at the office (on CentOS 5.1 x86_64). I've looked at some of the things in 3.0, but there's nothing there that I needed yet.

Hopefully you have some way to track changes in /etc/nagios (FSVS is what we use), because it will make your life much easier to have an audit trail.

We created sub-folders under /etc/nagios to hold the various types of entities. For example, we have:

/etc/nagios/commands /etc/nagios/contacts /etc/nagios/contactgroups /etc/nagios/hosts-switches /etc/nagios/hosts-dmz /etc/nagios/hosts-servers /etc/nagios/hosts-lan /etc/nagios/templates-hosts /etc/nagios/templates-services

We then broke individual elements out of the default massive configuration folder into individual .cfg files. For example, we chose to create individual files for each contact rather the putting them all in a single file. So far it works well, it's a lot easier to get a feel for what users have been defined, what hosts are defined, what the templates are. Because when I look in templates-services, I see from the directory listing that I have service templates named X, Y and Z (without having to open up the file to look).

We currently put service checks for individual hosts in the same configuration file as the host. So you will have the following definitions in a typical host file (until you get into templating):

define host{ define hostextinfo{ define service{ define service{ ...

Any plugins that we wrote ourself, we put under a separate folder. Which keeps them separate from

/usr/local/lib64/nagios-plugins/

Basically, start small, track your changes, and plan on refactoring it in week #2 after you start monitoring about a dozen hosts. Stay away from advanced things like escalation, monitoring things like disk space on remote servers, or the like until you get the basics working.

Oh, and SELinux will probably get in your way. So you'll need to play with audit2allow to create supplemental policy to give Nagios additional permissions. (Which may have been due to PEBKAC issues on my end - I plan on going back and looking at labeling and figuring out what I mislabeled.)

I think that's the majority of the issues that we dealt with in the past 2 weeks. We're now in fine-tuning mode and getting ready to start monitoring remote services next week.