NRPE on Scientific Linux

[jdelpeso]

I am evaluating Nagios XI to be implemented in our Computer Centre in the near future.
The first problem I have encountered is that it does not seem to support Scientific Linux
machines. In particular, the NRPE agent on a client cannot be installed if the client machine
is running Scientific Linux. I find it weird since Scientific Linux is derived from Red Hat Linux
and is extended across many computer centres all over the world. In particular, all our machines
in the computer center run Scientific Linux. Our computer centre is one of the nodes of the
Large Hadron Collider computing GRID.

Do you foresee to support Scientific Linux in the short-term ?
Is there a way to make it work at present ?

[tmcdonald]

What specific issues have you run into getting NRPE to run on SL?

As far as our commercial product Nagios XI goes, it needs to run on CentOS or RHEL v5/6/7 with a minimal install (no GUI).

[jdelpeso]

The test bench consists of two machines: the Nagios XI server (Server) and a client machine to be monitored (Client).
I have done three tests:
1.- NRPE-agent downloaded from Nagios XI on a SL Client. The "fullinstall" does not work. This is the error message:
"Scientific is not currently supported. Please use either Red Hat, CentOS, Oracle Linux, SUSE Enterprise, OpenSUSE, Ubuntu, or Debian."
2.- NRPE-agent downloaded from Nagios XI on a CentOS Client. The "fullinstall" works fine. The monitoring of this
Client by the Server works perfectly.
3.- NRPE downloaded from Nagios Core (free version) on a SL Client. The installation/configuration works perfectly on the SL client.
The Nagios XI monitoring does not monitor such machine properly.
For example, the command line send to check the /home mounting point does not work.
It seems that the place of the parameter -p matters!
check_command check_nrpe!check_disk!-a '-w 20% -c 10% -p /home'
it does not work, although is the command line build by Nagios XI to do the check.
check_command check_nrpe!check_disk!-a '-p /home -w 20% -c 10%'
it works properly
Hence, there is a kind of incompatibility between the Nagios XI server and a SL machine running nrpe from Nagios free version.

[abrist]

You should be able to alter the ./fullinstall script to install of SL (as it is a RHEL derivative).
Edit line number #72 from:

Code: Select all

if [ "$distro" == "CentOS" ] || [ "$distro" == "RedHatEnterpriseServer" ]; then

To:

Code: Select all

if [ "$distro" == "CentOS" ] || [ "$distro" == "RedHatEnterpriseServer" ] || [ "$distro" == "Scientific" ]; then

Save out and then try rerunning the ./fullinstall script.

[jdalrymple]

I'm seeing a bit different stuff than abrist - this in 2014r2.6

The quick hack(s):

Line 15 of init.sh

Code: Select all

xivar distro  "CentOS"

And line 45 of 0-repos

Code: Select all

        if [ "$dist" = el6 ] && ! rpm -q sl-release && ! rhn-channel -l | grep -q rhel-.\*-server-optional-6 ; then

Those 2 changes got me a working system - see attachment.

I'll make a note to see about getting the installer updated appropriately. Seems like something we should be able to tackle pretty easily, especially if it helps out the LHC!!!