fix(container): update image ghcr.io/tanguille/sglang-rdna4 (v0.5.15 ➔ v0.5.16)

[renovate]

This PR contains the following updates:

Package	Update	Change
ghcr.io/tanguille/sglang-rdna4	patch	v0.5.15-gfx1201 → v0.5.16-gfx1201

---

Configuration

📅 Schedule: (in timezone Europe/Brussels)

• Branch creation
  • At any time (no schedule defined)
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about these updates again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

Summary by CodeRabbit

• Chores
  • Updated the SGLang runtime used by the Qwen 3.6 27B inference service to a newer version.
  • No other inference service configuration changes.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 740f5dba-36b0-43f9-a94a-1e30b59d3133

📥 Commits

Reviewing files that changed from the base of the PR and between 3efff2a and 7b459b1.

📒 Files selected for processing (1)

• kubernetes/apps/ai/llmkube/models/qwen36-27b-sglang.yaml

---

📝 Walkthrough

Walkthrough

The Qwen36 SGLang InferenceService runtime image is updated from v0.5.15-gfx1201 to v0.5.16-gfx1201 with its corresponding pinned digest.

Changes

SGLang image update

Layer / File(s)	Summary
Update pinned SGLang runtime image kubernetes/apps/ai/llmkube/models/qwen36-27b-sglang.yaml	The InferenceService container image reference is updated to the v0.5.16-gfx1201 digest.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Possibly related PRs

• Tanguille/cluster#4219: Builds or publishes the corresponding v0.5.16-gfx1201 SGLang image.
• Tanguille/cluster#4167: Updates the same SGLang runtime image in the same manifest.
• Tanguille/cluster#4198: Updates the same pinned SGLang image reference.

Suggested reviewers: tanguille

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch renovate/ghcr.io-tanguille-sglang-rdna4-0.x

---

_{Comment @coderabbitai help to get the list of available commands.}

[tanguille-cluster]

@@ spec.image @@
# inference.llmkube.dev/v1alpha1/InferenceService/ai/qwen36-27b
! ± value change
- ghcr.io/tanguille/sglang-rdna4:v0.5.15-gfx1201@sha256:72d934d335529934bf73cd480b728cd08ae61178c9204dc37a025302b31fd61e
+ ghcr.io/tanguille/sglang-rdna4:v0.5.16-gfx1201@sha256:37bfc2e96b74d6f1b5024646f860db4053e621eb847c6691577badcb9752f461

[github-actions]

AI Automated Review

Analysis engine: qwen-3.6-fast@http://litellm.ai.svc.cluster.local/v1 (openai)

---

Recommendation: Request Changes

The image bump from v0.5.15-gfx1201 to v0.5.16-gfx1201 is correct, but the SGLANG_HICACHE_FILE_BACKEND_STORAGE_DIR environment variable still references the old version suffix (sglang-v0.5.15_awq-f541031d). The manifest's own comment explicitly states this suffix MUST be bumped with the image digest or AWQ revision to prevent stale KV pages from being served. This will cause incorrect cache hits after the upgrade.

---

Change-by-Change Findings

kubernetes/apps/ai/llmkube/models/qwen36-27b-sglang.yaml

• [major] HiCache storage directory not updated for new image version (line 98)
  The env var SGLANG_HICACHE_FILE_BACKEND_STORAGE_DIR remains set to /hicache/sglang-v0.5.15_awq-f541031d. The inline comment on that line reads:

  Page names encode neither the weights revision nor the engine, so this suffix MUST be bumped with the image digest above or the AWQ revision, or stale KV pages get served.
  Update the value to /hicache/sglang-v0.5.16_awq-f541031d to match the new image tag.

---

Standards Compliance

• The PR title follows Conventional Commits (fix(container): update image ...).
• Only one file changed; no shell scripts or SOPS secrets touched.
• No other repository conventions appear violated by the diff itself.

Unknowns or Needs Verification

• Whether v0.5.16-gfx1201 introduces breaking behavioral changes beyond what the HiCache suffix handles. The fork’s release notes are not publicly indexed; manual validation post-merge is recommended.