Improve ManagedNodeGroup when creating a LaunchTemplate

[JustASquid]

Hello!

• Vote on this issue by adding a 👍 reaction
• If you want to implement this feature, comment to let us know (we'll work with you on design, scheduling, etc.)

Issue details

In my current use case, I'm trying to implement a Managed Node Group with GPU support. Out of the box this isn't possible, however with a few small tweaks it should be entirely possible.

In addition, during my investigation I noticed a bug where it's not possible to specify the disk size of a managed node group when a custom launch template is created. This is because AWS requires the disk configuration to happen on the launch template, not the diskSize argument in this case.

My suggested implementation is in the PR #1225

Affected area/feature

ManagedNodeGroup

[t0yv0]

Thanks for looking into this @JustASquid ! At a first glance it looks helpful (CC @flostadler) - my team can help out shepherding the changes through the release process and getting the test cases passing on our CI; one quick request I have is can we split the two issues - one for Managed Node Group with GPU support and one for the diskSize bug, much appreciated!

[JustASquid]

Cheers @t0yv0 - I have split out the issue and the PR (#1229).

[flostadler]

This should be resolved once https://github.com/pulumi/pulumi-eks/pull/1339 lands. For now it'll first go into a feature branch dedicated for AL2023 & Bottlerocket support, but that one will land in the main branch within the next ~2 weeks

[flostadler]

This was fixed in v3 of the provider. Our automation just missed to close this issue out!