cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

PaaS Host memory detection differences (OpenShift Dedicated example)

Jon_Griffiths
Dynatrace Organizer
Dynatrace Organizer

Hi,

I was hoping someone could explain where the host metrics (especially memory) were coming from when deploying PaaS agents.

For OpenShift, RedHat provides host metrics where it would appear that the memory is almost always close to being full, whereas Dynatrace shows the host being half empty. Why is this? Where are we pulling our information from and how does that compare to RedHats?

For a specific example:
On one machine Dynatrace reports:
Total: 15.3GB, Amount Free: 7.7GB, Amount Used: 7.6GB
Whereas RedHat portal reports:
Total: 16GB, Amount Free: 1.5728GB, Amount Used: 14.4272GB

1 REPLY 1

michael_kopp
Dynatrace Pro
Dynatrace Pro

Dynatrace reports "available" memory and not "used". On a Linux host memory is always "full" simply because everything that is not used by processes is used for file caches and such.

Thus if u look at used memory on a linux box it will almost always be full. If u use the command "free" instead it will tell you how much of that is due to caches and what is "available" to be used by processes. The system will always clear out those caches if it is needed by processes.

So the 7.6GB that Dynatrace reports is the memory that is still used after you subtract the caches.

This has nothing to do with "PaaS" agents (which btw are not an official term ;), but is the case in general.

hope this helps