eks-distro-build-tooling
eks-distro-build-tooling copied to clipboard
aws
Break builder-base into multiple stages and layers
Issue #, if available:
Description of changes:
This breaks install.sh into separate scripts, and layers in the dockerfile, per component included in the builder-base. The end result is a mostly the same image, which a dozen layers vs 1 giant layer. Using this approach allows rebuilds locally to be a TON more manageable and it allows us to reuse the cache between different builds of builder-base, thus reducing build times even more.
Key changes:
- rust has been removed from the final builder base, install only tuftool is incldued
- go 1.14 has been removed, the final project we built 1.14 with was removed from eks-a not too long ago
- the instal golang script is currently not broken up by version, but as a follow up will be once @danbudris's work on custom golang builds is complete
I have diff'd the old image file list with a new image I built to confirm the only changes are the above expected ones.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.
/lgtm /approve /hold
Awesome work @jaxesn, really excited for this! One more abstraction to consider would be to extract the common wget -> shasum check -> extract and move to PATH steps to a common function and use it where it applies, to reduce the duplication in scripts. Not sure if you want to do this in this PR or want to do it as a follow-up, so I'm approving and put it on hold to merge at your discretion.
[APPROVALNOTIFIER] This PR is APPROVED
This pull-request has been approved by: abhay-krishna, jaxesn
The full list of commands accepted by this bot can be found here.
The pull request process is described here
- ~~OWNERS~~ [abhay-krishna,jaxesn]
Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment
I had the same thought. I'm going to leave that for the future. One reason I didn't jump to that was to avoid introducing too many common scripts that could trigger rebuilds of every layer and instead optd for duplication to keep them as independent as possible. Worth considering more tho for sure!