Make parallel build memory threshold configurable

[saschagrunert]

What type of PR is this?

/kind cleanup

What this PR does / why we need it:

The amount of memory required to build binaries in parallel is right now
set to 40GiB. We now make this variable to be able to build artifacts in
parallel even with a lower amount of memory.

This enables SIG Release to speed-up the build time drastically in
Google Cloud Build (GCB).

Which issue(s) this PR fixes:

None

Special notes for your reviewer:

cc @kubernetes/release-engineering
/sig release

Does this PR introduce a user-facing change?:

None

Additional documentation e.g., KEPs (Kubernetes Enhancement Proposals), usage docs, etc.:

None

[cpanato]

/lgtm

[xmudrii]

/lgtm

[saschagrunert]

/test pull-kubernetes-e2e-kind

[puerco]

/lgtm

[hasheddan]

/lgtm

[saschagrunert]

@kubernetes/milestone-maintainers can we get this over the finishing line?

[ehashman]

@saschagrunert that's not the issue, you're missing an approver:

kubernetes/hack/OWNERS

Lines 18 to 31 in 6d43e2b

	- bentheelder
	- cblecker
	- deads2k
	- dims # for local-up-cluster related files
	- fejta
	- lavalamp
	- liggitt
	- pwittrock
	- SataQiu
	- soltysh # for sig-cli related stuff
	- spiffxp
	- sttts
	- vishh
	- mikedanese

go chase one of these fine folks down :)

[cblecker]

/hold

So this variable is designed to be used as a deciding factor where if you have more than this amount of memory, you can build in parallel.. and below it, we don't because you experience errors during the build. I'm not clear on how making this configurable changes this.

I'd be okay with is making it configurable in general, but can we expand the comment to provide more context on what this variable does and why?

[cblecker]

This comment doesn't make sense anymore if this isn't hard coded

[saschagrunert]

Added a note that it can be overwritten at our own risk, whereas it still defaults to 40GB.

[saschagrunert]

@cblecker do you think you can give this another look? :)

[saschagrunert]

PING @fejta @deads2k @lavalamp may I ask you for an approval?

[BenTheElder]

If we're able to lower this, perhaps the default is higher than it needs to be?
I'd expect this to change over time with go releases, kubernetes changes ...

[saschagrunert]

If we're able to lower this, perhaps the default is higher than it needs to be?
I'd expect this to change over time with go releases, kubernetes changes ...

Yeah probably, this was just a test coming from our GCB instance. I was just a bit conservative in changing the default because I did not really know which impact it would have on existing build systems using older go versions. WDYT?

[justaugustus]

/area release-eng
/triage accepted

[BenTheElder]

Yeah probably, this was just a test coming from our GCB instance. I was just a bit conservative in changing the default because I did not really know which impact it would have on existing build systems using older go versions. WDYT?

Your GCB instance is using newer than k/k? Or is the concern about backporting to older branches?

Ideally we should have a reasonable default on the current branch, which will evolve with time?

[saschagrunert]

Yeah probably, this was just a test coming from our GCB instance. I was just a bit conservative in changing the default because I did not really know which impact it would have on existing build systems using older go versions. WDYT?

Your GCB instance is using newer than k/k? Or is the concern about backporting to older branches?

Yeah it was mainly about the backports.

Ideally we should have a reasonable default on the current branch, which will evolve with time?

I think we could play around with the value together with CI signal as follow-up.

[saschagrunert]

@BenTheElder do you think we can move forward with this PR and benchmark the actual required threshold after we made it configurable?

[spiffxp]

/approve
/lgtm
/hold cancel

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: cpanato, hasheddan, saschagrunert, spiffxp, xmudrii

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• hack/OWNERS [spiffxp]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[BenTheElder]

Sorry E_TOO_MANY_PRS. Looks like Aaron got it 😅

[spiffxp]

/test pull-kubernetes-unit
prow's status-reconciler should have triggered this but didn't, not sure why