Can we modify the way Metricbeat Claculates the System metric fighures

(Sunil Chaudhari) #1

I have configured MB on linux machine to read system metric.
However figures are different that the commands? When I say free -m on command line, t gives different figures than that of metricbeat.

How to validate output of metricbeat?
What are the commands internally it uses?
Can we modify those commands?

Sunil Chaudhari.

(Pier-Hugues Pellerin) #2

We are using a system call to retrieve this information.

I think we send the bytes values to elasticsearch, and in your case free -m returns the values in mebibytes

How different are the numbers it is just a scale difference?

(Sunil Chaudhari) #3

Hi Pier,
Thanks for reply.

There is scale different problem as well as explained below.
There are multiple hosts i have configured. In Combine line graph it shows one system to some Y1 level. See image 1
And when I click on that hosts, it goes down to some lower level, which seems to be its original one. Why does it behave like this?
See the host with green color.



(Sunil Chaudhari) #4

About calculation difference, We need the figures as below for the memory:
How we calculate free memory, is (free memory) + (buffer/cacche memory)
How we calculate memory utilization is equal to List all the running services and add the total percentage usage of each service
via (pidstat -rl)

How can we modify metricbeat to align to above calculations output.


(Pier-Hugues Pellerin) #5

Are theses the default dashboard that we ship with metricbeat?

(Sunil Chaudhari) #6

Yes. These are default dashboards and visualization too.

(system) #7

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.