Re: Node Exporter

[Yasmine Mbarek]

Hello , 

I have a tiny problem with node exporter. If you can help me I will be very grateful .

So my node exporter implemented in my parc of machines , for some machine it works fine and returns all metrics values but in other machine it returns everything but "RAM Used" , the difference between the working machines and the others is that : on the machines that RAM used is working fine , the OS is redhat 7 

the rest of machines Redhat 6 everything is working but the RAM used returns NO DATA 

Is there any  explanation ??

Thank you .

ᐧ

Le mar. 16 juin 2020 à 16:55, Yasmine Mbarek <bm.ya...@gmail.com> a écrit :

Hello , 

I have a tiny problem with node exporter. If you can help me I will be very grateful .

So my node exporter implemented in my parc of machines , for some machine it works fine and returns all metrics values but in other machine it returns everything but "RAM Used" , the difference between the working machines and the others is that : on the machines that RAM used is working fine , the OS is redhat 7 

the rest of machines Redhat 6 everything is working but the RAM used returns NO DATA 

Is there any  explanation ??

Thank you .

--

أطيب التحيات  / Cordialement / Best regards,
----------------------

Yasmine MBAREK

Emails : yasmine...@tek-up.de

              bm.ya...@gmail.com

Phone : +216 52 94 61 38 

ᐧ

--

أطيب التحيات  / Cordialement / Best regards,
----------------------

Yasmine MBAREK

Emails : yasmine...@tek-up.de

              bm.ya...@gmail.com

Phone : +216 52 94 61 38 

[Christian Hoffmann]

Hi,

On 6/17/20 4:44 PM, Yasmine Mbarek wrote:
> I have a tiny problem with node exporter. If you can help me I will be
> very grateful .
> So my node exporter implemented in my parc of machines , for some
> machine it works fine and returns all metrics values but in other
> machine it returns everything but "RAM Used" , the difference between
> the working machines and the others is that : on the machines that RAM
> used is working fine , the OS is redhat 7 
> the rest of machines Redhat 6 everything is working but the RAM used
> returns NO DATA 
> Is there any  explanation ??

I assume you are using Grafana with some kind of node_exporter dashboard?

You would have to look into the actual queries to see what's causing this.

My guess: The dashboard most likely uses the MemAvailable metric (which
comes from /proc/meminfo). This is exposed by the Linux kernel, but only
after some specific version. RHEL6 does not expose it.

There are ways to calculate something similar for RHEL6 (the "free"
command line tool has some logic for this).

I think GitLab had a public Prometheus config with a recording rule,
maybe it was this one:

https://gitlab.com/gitlab-org/gitlab-foss/-/commit/e91c7469ad0be5f429548d4142ca93c17ec9e71e

You would have to set up such a recording rule and would have to modify
your dashboard accordingly.

As an alternative: Try talking your administrators into abandoning
RHEL6. It'll be out of support at the end of the year anyway, IIRC. :)

Kind regards,
Christian

[Yasmine Mbarek]

Thank you very much , i will study your solution and i hope it can be beneficial :)))

[Ben Kochie]

Yes, we use a recording rules to represent node memory utilization.

https://gitlab.com/gitlab-com/runbooks/-/blob/1682a5632f3eaf0548dfa8277a421de2aff24245/rules/node.yml#L91-106

--
You received this message because you are subscribed to the Google Groups "Prometheus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/6fbaa443-48f3-1bd0-962d-490e4cde4b80%40hoffmann-christian.info.

[Yasmine Mbarek]

Sorry but where to add all of these modifications ? the node_exporter file on my remote host is a binary file ?!

To unsubscribe from this group and stop receiving emails from it, send an email to promethe...@googlegroups.com.