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

Infrastructure Monitoring

kisu_pyo
Participant

Hi,

I have managed incidents using Dynatrace and made SMS send to each person in charge of each system.

These days my customers got SMS saying that used OS memory is over 95% of total memory capacity, but after checking their system, they said the system has lots of usable memory, and they don't think the system has any problems.

When I check Dynatrace, it is true that used memory occupies 96% of total memory. (7.7 GB of 8GB are used ).

To verify this phenomena, I checked my system with dynatrace. My system is based on linux, and I figured out 20GB of 60GB are used using "free -t" command. But dynatrace said only 5GB of 60GB are used.

I think there is some kind of misunderstanding about "memory" between physical system and dynatrace data.

Do you know what it is?

5 REPLIES 5

Babar_Qayyum
Leader

Hello Kisu,

DynaTrace has its own thresholds by default so you can change as per your need or match with the other infrastructure monitoring.

You can configure on the host and group level.

Settings >> DynaTrace Server >> Infrastructure

Regards,

Babar

Yes, with this guide, I can adjust thresholds and then the incidents won't happen. But what I want to know is the difference between the value from physical system and that from dynatrace. Or, I want to know whether there's another concept of infrastructure monitored value from dynatrace.

Dear Kisu,

A monitored process can have a different settings than the host. Belwo screenshot is as a reference where you can see the physical memory for the host and the different memory settings for the monitored process.

For further understanding about host monitoring, please check the below link.

https://community.dynatrace.com/community/display/...

Regards,

Babar

seongwoon_na1
Dynatrace Guide
Dynatrace Guide

I am currently on training in Linz, so response may be late.


Joe_Hoffman
Dynatrace Leader
Dynatrace Leader

Kisu, I don't believe your issue is the thresholds. Your issue is that AppMon is reporting different values than you're familiar with using your own command line tools. This can happen as a result of different interpretations of what is considered "Free" and "Used" and "Committed" memory.

AppMon uses "/proc/meminfo" to extract memory metrics from your Linux machine. The following definitions are what's provided by the Linux core team:

Available: The available memory is defined as: free+Cached+Buffers
Free: 'Free' is actually the metric 'Usable Memory', which again is free+buffers+cached.
Used: 'Used' is total physical memory-usable memory.
Commited:There are two forms of used memory. with and without buffered ram. Used is really used by programs and can not be freed without swapping. Committed is including buffers and caches which are freed if needed.

Based on these definitions does this help clarify the values you're seeing? Perhaps you can inspect /proc/meminfo on your system and see whether they match.