To use pods to maximize utilization of node resources instead of using that node as VM

General Discussions
1

Use case:
I run tests on VMs of having 6GB RAM and 4 core processors . I found that during test run load average never crosses 0.4 or 0.5 so that means my test can run lesser core and on lesser RAM so I thought why not create a pod with after setting some service and request limit for CPU and RAM and set number of pods limit for a node and try to squeeze in as many as many pods I can irrespective of node size considering I do not overdo it which led to memory leak or killing of container.

Doubt:
Is this doable ?.I need pod to run tests and not to launch any webapp or load balance any application.I just want to increase utilization of VM so that I can save money and run more tests in parallel rather than on multiple VMs of average size and getting charged for all.I will be getting instance from Cloud so I can assign lower size VM to cater my needs but I am looking for generic solution where I can fit in more pods in VM of any size as Kubernetes cluster by just modifying some param based on size of VM I get.

Thanks for reading it.Please suggest.

2

Kindly suggest.

3

Have you tested it out? This document explains managing resources for containers. Keep in mind the resources are enforced by the container runtime and the behavior may vary there.

4

Oh and there’s a max pod limit per node. It’s a kublet setting (--max-pods).

There is also considerations for IP address limitations to consider as well. IP allocation is finite.