Request for ability to control memory/cpu resource allocated to nodes

[zephinzer]

What would you like to be added: Ability to specify memory/cpu values for each node

Why is this needed: Having this would enable localised testing of workload behaviour mimicing a real cluster's resources (example question that can be answered: does OOM happen given we are using 120% memory based on limits but 80% based on requests?)

Haven't dug into how kind works so I'm not too sure how complex this will be but the Docker daemon seems to allow for specifying of resources so it could be adding configuration during this step?

If it's in line with kind's roadmap, I can help with this-

[zephinzer]

Just found #877, which seems to be a duplicate, feel free to close-

[BenTheElder]

This is unfortunately trickier than you might hope to do correctly at the moment and not something I personally can justify focusing on at the moment. If someone proposes a workable plan in #877 I will do my best to help see it through. I think most/all? of the kind maintainers are currently burning all cycles maintaining the project and doing their "real" work on other projects 😅