test chat completions api in e2e case

[delavet]

Adds /chat/completions API e2e test case.

Fixes: #814

[netlify]

✅ Deploy Preview for gateway-api-inference-extension ready!

Name	Link
🔨 Latest commit	451fea1
🔍 Latest deploy log	https://app.netlify.com/projects/gateway-api-inference-extension/deploys/683e664a2113f9000855449b
😎 Deploy Preview	https://deploy-preview-868--gateway-api-inference-extension.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[k8s-ci-robot]

Hi @delavet. Thanks for your PR.

I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[nirrozenbaum]

/ok-to-test

[nirrozenbaum]

@delavet does it make sense to separate different tests into different ginkgo.It statements?
I'd expect moving some parts to the BeforeEach section, e.g., this part:

ginkgo.By("Creating an InferenceModel resource")
infModel := newInferenceModel(nsName)	
gomega.Expect(cli.Create(ctx, infModel)).To(gomega.Succeed())	

ginkgo.By("Ensuring the InferenceModel resource exists in the namespace")
gomega.Eventually(func() error {
	return cli.Get(ctx, types.NamespacedName{Namespace: infModel.Namespace, Name: infModel.Name}, infModel)	}, 
existsTimeout, interval).Should(gomega.Succeed())

I'd also expect creating some helper functions to extract the common code.
I think the e2e tests should include (each in a different ginkgo.It statement):

• completions with prompt
• chat completions with a single [role, message].
• chat completions with multiple [role, message] entries.

[delavet]

@delavet does it make sense to separate different tests into different ginkgo.It statements? I'd expect moving some parts to the BeforeEach section, e.g., this part:

ginkgo.By("Creating an InferenceModel resource")
infModel := newInferenceModel(nsName)	
gomega.Expect(cli.Create(ctx, infModel)).To(gomega.Succeed())	

ginkgo.By("Ensuring the InferenceModel resource exists in the namespace")
gomega.Eventually(func() error {
	return cli.Get(ctx, types.NamespacedName{Namespace: infModel.Namespace, Name: infModel.Name}, infModel)	}, 
existsTimeout, interval).Should(gomega.Succeed())

I'd also expect creating some helper functions to extract the common code. I think the e2e tests should include (each in a different ginkgo.It statement):

• completions with prompt
• chat completions with a single [role, message].
• chat completions with multiple [role, message] entries.

This does make more sense! I will try refactoring this ASAP.

[danehans]

@delavet does it make sense to separate different tests into different ginkgo.It statements?

Another option is to move the ginkgo.By() statement within the range and include the name of the API endpoint so a user can see that each endpoint is being tested?

[delavet]

@nirrozenbaum @danehans I have refined this PR, and now the e2e tests include coverage for both /completion and /chat/completions. The tests for /chat/completions encompass two scenarios: single-turn chat and multi-turn chat. For the single-turn chat, the same prompt used in /completion, "Write as if you were a critic: San Francisco," is applied. Regarding the organization of the tests, I finally decided to use a single It block that contains multiple By blocks through a loop. This approach appears more semantically coherent.

[danehans]

@delavet CI is failing due to #901. Please rebase to include #902.

[nirrozenbaum]

/retest

[danehans]

Changes verified:

$ make test-e2e
...
  STEP: Creating an InferenceModel resource @ 06/03/25 09:04:06.069
  STEP: Ensuring the InferenceModel resource exists in the namespace @ 06/03/25 09:04:06.074
  STEP: Verifying connectivity through the inference extension with /completions api and prompt/messages: Write as if you were a critic: San Francisco @ 06/03/25 09:04:06.082
  STEP: Verifying connectivity through the inference extension with /chat/completions api and prompt/messages: [{"role": "user", "content": "Write as if you were a critic: San Francisco"}] @ 06/03/25 09:04:06.275
  STEP: Verifying connectivity through the inference extension with /chat/completions api and prompt/messages: [{"role": "user", "content": "Write as if you were a critic: San Francisco"},{"role": "assistant", "content": "Okay, let's see..."},{"role": "user", "content": "Now summarize your thoughts."}] @ 06/03/25 09:04:06.369
  STEP: Deleting the InferenceModel test resource. @ 06/03/25 09:04:06.454
• [0.394 seconds]
------------------------------
[AfterSuite]
/Users/solo-system-dhansen/go/src/sigs.k8s.io/gateway-api-inference-extension/test/e2e/epp/e2e_suite_test.go:140
  STEP: Performing global cleanup @ 06/03/25 09:04:06.46
[AfterSuite] PASSED [0.203 seconds]
------------------------------

Ran 1 of 1 Specs in 79.008 seconds
SUCCESS! -- 1 Passed | 0 Failed | 0 Pending | 0 Skipped
--- PASS: TestAPIs (79.01s)
PASS
ok  	sigs.k8s.io/gateway-api-inference-extension/test/e2e/epp	79.036s

/lgtm
/approve

Thanks @delavet for your help!

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: danehans, delavet

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:

• OWNERS [danehans]

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