Nutanix Monitoring

I made an up to date version of my extended Nutanix monitoring.
It brings Nutanix on the “same” level as the ESX monitoring, i hope so :slight_smile:

This version should be the last stand alone extension to this included agent and checks.
I migrated all legacy plugins to the new API and asked for how-to submit the whole bunch of data to the CMK github. A PR with around 50-60 files is not liked by the Munich team i think.

The 2.1 package should also work with 2.0 - if someone can test it, it would be helpful.
The packages marked with 2.2 and complete please ignore.

9 Likes

Update - i got information that the 2.1 package is not completely compatible with 2.0
I will have a look at this problem the next days.
The problem are with the check parameters as far as i know.

2 Likes

Updated the 2.0 package here

Install is working without problem, quick function test was also without problem.

1 Like

Thanks Andreas for these great checks. They’re much more helpful than that what comes with checkmk itself. I hope tribe29 will include this in future checkmk releases.

@andreas-doehler I get:
$ cmk -D foo.bar.com
Failed to auto-migrate legacy plugin to check plugin: prism_alerts
Please refer to Werk 10601 for more information.
Failed to auto-migrate legacy plugin to check plugin: prism_containers
Please refer to Werk 10601 for more information.
Failed to auto-migrate legacy plugin to check plugin: prism_info
Please refer to Werk 10601 for more information.
Failed to auto-migrate legacy plugin to check plugin: prism_storage_pools
Please refer to Werk 10601 for more information.

Do we need to remove the existing cmk nutanix integration before?

Normally this message is only informational.
What happens on your host if you do a “cmk -vvI foo.bar.com” or a “cmk -vvn foo.bar.com”?

You need to manually remove the old plugins only if you want to use the “–debug” switch.

I get

  Source: SourceType.HOST/FetcherType.PROGRAM
  -> Not adding sections: Agent exited with code 1: Caught unhandled HTTPError(HTTP Error 404: Not Found) in /omd/sites/xxx/lib/python3/cmk/special_agents/utils/agent_common.py:135
  Source: SourceType.HOST/FetcherType.PIGGYBACK

You can execute the agent manually. The command is shown with an “cmk -D hostname”.
It looks more like, that there is not the right URL requested.

Program: /omd/sites/xxx/share/check_mk/agents/special/agent_prism '--server' 'x.x.x.x' '--port' '9440' '--username' 'foo' '--password' 'bar'
Servername, Port, User and Password is ok, but I’m wondering why it seems to use the builtin agent_prism?

Ok that is strange - what package of the prism agent from my github do you have installed?
If it is CMK 2.1 than it can be ok
The shown path is only a wrapper and it calls then the right function from ~/local/lib/check_mk/…

nutanix_prism-3.0-2.1.mkp We’re running 2.1.0p14

What happens if you execute the special agent command manually - some more output or only the same HTTP 404?

Found one error inside the special agent and will fix it.

Updated package nutanix_prism-3.1-2.1.mkp is available on github.
I hope this was the last copy & paste error.

Now also the PR on Github is made to integrate all these fancy new checks inside the base system.
We can only cross the fingers and pray :smiley:

1 Like

Hi Andreas
i installed your package for 2.0 (v 3.3) on a 2.0.0p27 and have it run against a Nutanix prism 5.20.3. (i know its not the most recent one but the only one i can test it now).
Where the builtin checks work (1st screenshot), the checks of your MKP throw an exception (2nd screenshot):
grafik
grafik

This is what the crash report looks like of one of the container checks:

And this is the one of the discovery:

Let me know if you need more information on this (via DM then?)

BR
Thomas

I got the same crash some days before.
It was a problem with the API versions of Nutanix - some calls need API v1 and most other calls v2.

New version online - 3.4
I hope this was the last problem between API v1 and v2 - but API version 3 is waiting :smiley:

Andreas, superb work, as always:

Now i have one crash left:

I suspect its caused by the hostname (which is its IP)?

BR
Thomas

For this i think i need the agent output.
It looks a little bit strange. As it means this host has no memory.

Sent a DM to you :slight_smile:

BR
Thomas