What are the possible causes of memory leaks in Kubernetes .Net applications?

Question

ubuntu

Good day!
There are applications written in .Net core. Running in Docker, they work fine. But on another server, being launched via Kubernetes, such strange behavior is observed: after rebooting the dev server, the total memory used is about 3 GB out of 6. Over time, when the deployment occurs, the memory used (namely, busy, not buffer / cache) grows , and grows until it takes all 6 GB, after which the OOM-killer and other joys of failure come up until the reboot. Restarting pods, cluster, docker does not help. The memory seems to have been eaten away by something. At the same time, if you sum up the memory occupied by applications, then the amount comes out in the region of 3 GB. Where can the "occupied" memory "leak"?
Kubernetes 1.17, Docker 19.03, kernel 4.18

Answer 1

Running in Docker, they work fine. But on another server, being run through Kubernetes

Same Dockerfile and docker version?

At the same time, if you sum up the memory occupied by applications, then the amount comes out in the region of 3 GB.

What are you monitoring?