High memory usage in Kubernetes charms

[Rui Vasconcelos]

We have been exploring the memory consumption of Kubeflow on top of Microk8s, as we have found Microk8s+Kubeflow bundle too big to fit into a laptop with 16GB of RAM running Mac or Windows.

In our Kubeflow memory analysis:
https://docs.google.com/spreadsheets/d/1lzZw2WEf-a2C9orQNjq0y_8WMJ9xbidwE7RTuU0vvTw/edit

we have identified that a big portion of the memory is consumed by juju agents, even when deploying workloads with a residual memory footprint.

[Pen Gale (pengale)]

Thank you for the detailed analyses!

I think that some of this may be addressed by ongoing work this cycle and next, both refactoring and combining Juju agents, and re-thinking some of how the k8s charms work.

[Canonical Juju QA Bot (juju-qa-bot)]

This bug has not been updated in 2 years, so we're marking it Low importance. If you believe this is incorrect, please update the importance.