Convox Community

CPU Utilization with K8s rack on Mac

Pretty consistent 30-40% cpu from Docker for me.

I reinstalled my rack fresh and there is a single app running a single container that has a CMD of ‘sleep’.

I enabled ‘Show system containers’ for K8s and I can see a few processes consistently using cpu:

15b80351b705        k8s_kube-controller-manager_kube-controller-manager-docker-for-desktop_kube-system_710c0d3b4f3cd8f2ee65068bcc49198e_0   4.12%               45.71MiB / 7.787GiB   0.57%               0B / 0B             56.9MB / 0B         9
73eb40ae6b25        k8s_etcd_etcd-docker-for-desktop_kube-system_3dfb2516a268bf8b0d43a783fad533f6_0                                         4.45%               55.34MiB / 7.787GiB   0.69%               0B / 0B             29.4MB / 10.7GB     17
db877dc70d50        k8s_kube-scheduler_kube-scheduler-docker-for-desktop_kube-system_b6155a27330304c86badfef38a6b483b_0                     1.72%               12.85MiB / 7.787GiB   0.16%               0B / 0B             27MB / 0B           8
d84c1d7b1cde        k8s_kube-apiserver_kube-apiserver-docker-for-desktop_kube-system_723b614df011d76edcaf6f999ad18b83_0                     4.67%               564MiB / 7.787GiB     7.07%               0B / 0B             66.8MB / 0B         10

It kind of seems like this is k8s and not convox, but I’m not sure. Is there anything that can be done about this, or are we just stuck with it?

Did a little googling, looks like we’re stuck with it:

Only 30%? Mine’s pretty consistently on 90% * CPU so if I allocate 2 cores to Docker for Mac its often at 180%.

I did see a slight improvement when I reset and moved to Docker Edge but it might just been the reset :man_shrugging:

I’m on Docker Edge and my utilization is generally fairly low but it definitely seems like they have a lot of room for improvement.