openshift agents memory limit metric

Document ID : KB000117364
Last Modified Date : 11/10/2018
Show Technical Document Details
Issue:
The memory limit metric reported for a pod by the OpenShift agent seems to reflect the whole system memory not what has been defined as a maximum for the pod
Environment:
APM 10.7 OpenShift on Linux
Cause:
Customer was just setting a variable value (MEMORY_LIMIT=2Gb) in the OS console and there was nothing behind that to actually affect the memory limit - this was verified by checking the memory limits using the docker stats command, so independent of APM
Resolution:
There are a couple of mechanisms you can use to define the memory limit in Openshift.
 - using the Openshift Console use the 'Edit Resource Limits' menu option in the Actions menu for the pod
 - Define the limit directly in the yml file e.g.

resources:
limits:
cpu: "0.4"
memory: "400Mi"
requests:
cpu: "0.04"
memory: "40Mi"


When these mechanisms are employed the OpenShift Agent reflects the  value define in OpenShift