Nutanix Monitoring

Ok bug with not connected hosts is fixed now :slight_smile:
v3.6 of the mkp for 2.0 and 2.1 is available - i hope that was the last edge case

1 Like

Works like a charm!
Thank you very much :slight_smile:

BR
Thomas

New package version available.

With some extra checks - contributed from @jwiederh
At the moment only 2.1 - i need some time to test and fix this for 2.0

4 Likes

Hey Andreas, do you also have a package for 2.2?

You can checkout the new 2.2 package.
Pay attention you need to uninstall/disable the old package before.
This is the result of the version number problem we are experience at the moment :frowning:

2 Likes

Hey,

I installed/enabled the plugin using mkp.
but i don’t understand how to use it in wato.

I added host gave it the ip of the cluster, used snmp cred, saved and went to services…
it shows checks of one of the CVM’s and not the cluster itself.

:frowning:

You need to use the IP of your Prism instance.
There is no SNMP used.
Please use the “Nutanix Prism” special agent rule for this host.

that was fast response, so first of all thank you!

Check the screenshot, am I at the right place?

I don’t see any field for IP/Host of Prism.

EDIT
I created new Host.
Then went back to this window and checked the “Explicit hosts” and selected the host I’ve created.
Then it worked.

Thank you!

Can I use it on Prism Central?

The problem with Prism Central is that it is not providing all the data needed as Prism Central don’t know all the old API functions if i remember it correctly.

Prism Central is more managing than monitoring.

Thank you. :pray: :pray: :pray: :pray: :pray:

Hey @andreas-doehler,

we have a large nutanix cluster and since some weeks (I’m not sure if since upgrade to 2.3 or later) it takes ~3 minutes to do the vm query. We opened a ticket at nutanix and they said that v1 API is deprecated and we should use v2 instead. And indeed we can see that with acli, which uses v2, it just takes a few seconds. So is it possible for you to do the vm query with api v2? It seems that you use v2 already for other checks.

Then they should also say how to get the information with v2 that are fetched with v1 inside the agent. That will not happen as v2 of the API is really broken in terms of data it delivers.

From inside the API documentation to v1

Available but deprecated; should not be used directly unless a v1-only action is required (e.g. virtual machine performance metrics)

The information in brackets are the important one.

At the moment i look at the API v4 version as this should combine v1, v2 and v3. With this v4 should also provide performance metrics for VMs.

v3 was a PrismCentral only API

@andreas-doehler

I updated my test instance today from 2.2 to 2.3.0p15
I noticed, that the value “ignore” in the ruleset “Nutanix VM State” is missing now. I want to get the VM states, but don’t care about them. With the mkp in CheckMK 2.2 I could set the state to “ignore” in the ruleset.

Is there maybe another way / value I have to set the rule to?
Thanks
Christoph

I also had this problem on a system last week. The problem here is the code submit to the main Github repo was over 1,5 years before now. In the mean time the 2.2 version got some bugfixes and as i must say - at the moment it is nearly impossible to get bugfixes into the main line back.
Good example is the Fritzbox fix that was automatically closed as the stupid CLA thingy is broken and then no one takes a look at valid commits.

@martin.hirschvogel this problem here with the broken Github things should be a high priority item.

I will try to make a PR for this fix that was made in the 2.2 version.

Hey @andreas-doehler,

sorry for the late reply. Nutanix confirmed what you said about the v2 API. Would it be an Option to use the Nutanix Plugin but disable the metrics gathering in an easy way?

Not with the current build of the agent. It needs to be extended with an option to select single sections or disable sections.

That would be great. Another thing that they suggested to us is to make some calls parallel and use pagination. We tried that with curl and got a significant improvement

I asked my colleague to take a look at the CLA bot in that specific case. We have started working on pull requests again.

1 Like