bump llm-d-x images to latest tags & improve ci

[nicole-lihui]

update llm-d-model-service tag version to v0.0.15
fix llm-d/llm-d-model-service#223

optimization makefile

• makefile supported auto bump-image-tags
• opt bump-modelservice-crd specify version & --depth=1

[nerdalert]

Hi @nicole-lihui appreciate the PR. I'm happy to close #342 for this. We are cutting a release this week and decomposing the charts and bump still need to keep this set of charts alive for now. Can you bump the Chart for CI again pls https://github.com/llm-d/llm-d-deployer/blob/main/charts/llm-d/Chart.yaml#L4

Odds are that version will get stomped on as a release is cut, so apologies now. Just wanted to warn you upfront it might be annoying since it's pretty much a release versioning cut 😅 Will test end to end with this today.

[nicole-lihui]

🙌 Chart version bumped! Let me know if there's anything else needed.

Thanks for the review @nerdalert

[nerdalert]

Hi @nicole-lihui I tested this and it LGTM. Ty! cc/ @Gregory-Pereira at your convenience take a peek.

[nicole-lihui]

Hi @nicole-lihui I tested this and it LGTM. Ty! cc/ @Gregory-Pereira at your convenience take a peek.

Hi @nerdalert,
this chart is ok right, but I noticed there's an issue with the workflow.

The "Test Latest Release" job doesn't actually pick up the new values when running helm upgrade.

I try use helm upgrade flags : --reset-values , this is incorrect methods

[nicole-lihui]

😟 ct default use --reuse-values

helm/chart-testing#525

Is there a way to work around this behavior in the workflow? I’m not quite sure about this part.

[nicole-lihui]

ct fixed in helm/chart-testing#742, version is v3.13.0

but helm action not released

[nerdalert]

Hi @nicole-lihui I've done some basic testing with base and p/d deployments and things looked good. I'd like to bump this Monday when we cut a release. Also wanted to make sure you have seen this which we are migrating to as this deprecates at some point for the decomposed approach here https://github.com/llm-d-incubation/llm-d-infra Ty!

[nicole-lihui]

https://github.com/llm-d-incubation/llm-d-infra

@nerdalert Appreciate you sharing the roadmap of llm-d deployment mode! I like this decompose setup in llm-d-infra -- the current large values can be a bit messy.

I'll stay tuned and would love to keep contributing!

[nerdalert]

LGTM thanks for the patience and effort here @nicole-lihui!