Differences in memory usage between heapster and os view for a build pod in Red Hat OpenShift Container Platform
Issue
- When running a build (
nodejs
with headlesschrome
for testing) we see lot's of the time, the build being killed due to out of memory condition. The problem is, thatoc adm top pod
is constantly reporting 128 MiB Memory usage for this particular build pod. The limits for the pod are set to 2 GiB for the memory. When checkingdocker stats
for the particular container it reports 1.8 GiB Memory usage. We therefore have a notable difference in memory usage betweendocker
andoc adm top pod
and are wondering where this is coming from. Also note, this only happens for build pods. Regular running pods are reporting correct values.
Environment
- Red Hat OpenShift Container Platform 3.x
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.