[WONTFIX]: Platform Engineer can limit memory used during staging using stagingRequirements.memoryMB

[tcdowney]

Blockers/Dependencies

No response

Background

As a Platform Engineer
I want to be able to set a memory limit for staging apps
So that I have more predictable utilization of my platform

Acceptance Criteria

GIVEN I have configured the stagingRequirements.memoryMB Helm value
WHEN I cf push an app
THEN I see that there is a memory resource limit on the kpack Image / Pod performing the build corresponding with the value I set

Dev Notes

• NOTE this property already exists in the Helm chart, it is just not being used. We should be able to specify it as a resource limit here.
• We want to set this as a limit instead of a request for now since a request may limit Pod placement too much. Eventually we'd want to support specifying the request on a per CFApp/CFBuild level since apps will have varying requirements here.
• Ephemeral disk story: stagingRequirements.diskMB

[tcdowney]

I left a comment with some thoughts on implementation the disk story that also apply to this one:
#2651 (comment)

[tcdowney]

This was effectively undone by #2685.

This feature was originally oriented around setting limits. After some discussion on a Slack thread and a Github issue we decided to undo that. The original CF installation properties were interpreted as a minimum resource request versus a hard limit so we made this value optional (no requests) and allow operators to specify it if they want to set that minimum.

CF for VMs uses a quota system to set operator-specified limits -- so that might be something worth exploring if we want to do this in the future. Given that we don't want to do this right now and that the implementation/UX should likely look very different, I am going to close this issue as "wontfix".