fix(customizer): change entrypoint of side car to avoid API spin up for LoRA and unwritable dirs

[soluwalana]

Summary

Related Issue

Changes

Type of Change

• Code change (feature, bug fix, or refactor)
• Code change with documentation updates
• Documentation only
• Contributor tooling or automation
• CI, build, or test infrastructure

Quality Gates

• Tests added or updated for changed behavior
• Existing tests cover changed behavior — justification:
• Tests not applicable — justification:
• Documentation updated for user-visible behavior
• Documentation not applicable — justification:

Verification

• Pull request title follows the repository's Conventional Commit format
• Every commit includes an appropriate Signed-off-by: trailer
• uv run pre-commit run -a passes, or any blocked checks are identified below
• Targeted tests pass, or tests are marked not applicable above
• No secrets, API keys, or credentials are included

Targeted validation:

Summary by CodeRabbit

• New Features

  • LoRA adapter sidecars now launch directly through the Python adapters module during model deployments.
  • Added configuration support for customizing the LoRA sidecar launch command.
  • Sidecar state, data, and configuration files now use the writable scratch volume for improved deployment compatibility.
  • Updated deployment defaults and validation to use the new launch method.

• Documentation

  • Updated the configuration reference with the new default sidecar command.

[github-actions]

🌿 Preview your docs: https://nvidia-preview-pr-1185-nvbug6573168-solu.docs.buildwithfern.com/nemo-platform

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 50b91e4c-51cc-4b06-ae09-594d972326db

📥 Commits

Reviewing files that changed from the base of the PR and between 502a651 and b50edec.

📒 Files selected for processing (2)

• services/core/models/src/nmp/core/models/controllers/backends/deployments_plugin/compiler.py
• services/core/models/tests/unit/controllers/backends/deployments_plugin/test_compiler.py

🚧 Files skipped from review as they are similar to previous changes (1)

• services/core/models/tests/unit/controllers/backends/deployments_plugin/test_compiler.py

---

📝 Walkthrough

Walkthrough

The LoRA adapters sidecar now launches through direct Python module execution. Deployment configuration, Helm values, Docker helpers, compiler tests, compiler environment setup, and setup documentation use the updated command and writable scratch-backed XDG directories.

Changes

LoRA sidecar command

Layer / File(s)	Summary
Command contract and deployment configuration services/core/models/src/.../deployments_plugin/config.py, k8s/helm/values.yaml, plugins/nemo-deployments/tests/unit/.../docker_helpers.py, docs/set-up/config-reference.mdx	The default, Helm values, Docker helper, and configuration reference use python -m nmp.core.models.sidecars.adapters.main.
Scratch-backed XDG directories services/core/models/src/.../deployments_plugin/compiler.py	The LoRA sidecar redirects XDG state, data, and config paths to the writable scratch volume.
Sidecar validation services/core/models/tests/unit/.../test_compiler.py	The compiler test checks the direct module command and validates all three scratch-backed XDG environment variables.

Possibly related PRs

• NVIDIA-NeMo/nemo-platform#663: Both PRs modify the LoRA sidecar command configuration.
• NVIDIA-NeMo/nemo-platform#798: Both PRs modify writable XDG directory handling for the LoRA sidecar.
• NVIDIA-NeMo/nemo-platform#870: Both PRs modify the Docker LoRA adapters sidecar command.

Suggested reviewers: tylersbray, mckornfield

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the entrypoint change and the fixes for API startup and unwritable directories.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch nvbug6573168/solu

---

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

[coderabbitai]

🧹 Nitpick comments (1)

services/core/models/tests/unit/controllers/backends/deployments_plugin/test_compiler.py (1)

362-362: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Assert the Docker command too.

The test compiles both runtimes, but Line 362 checks only the Kubernetes sidecar. Add an assertion for docker.server_config.containers[1].command.

Proposed assertion

     assert sidecar.command == ["python", "-m", "nmp.core.models.sidecars.adapters.main"]
+    assert docker.server_config.containers[1].command == [
+        "python",
+        "-m",
+        "nmp.core.models.sidecars.adapters.main",
+    ]

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In
`@services/core/models/tests/unit/controllers/backends/deployments_plugin/test_compiler.py`
at line 362, Add an assertion alongside the existing Kubernetes sidecar check in
the test covering both compiled runtimes, validating that
docker.server_config.containers[1].command matches the expected Python module
command.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In
`@services/core/models/tests/unit/controllers/backends/deployments_plugin/test_compiler.py`:
- Line 362: Add an assertion alongside the existing Kubernetes sidecar check in
the test covering both compiled runtimes, validating that
docker.server_config.containers[1].command matches the expected Python module
command.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: aca0f37a-751b-4f8f-895e-43110ea55e99

📥 Commits

Reviewing files that changed from the base of the PR and between 6ed5aa1 and 95b690f.

📒 Files selected for processing (5)

• docs/set-up/config-reference.mdx
• k8s/helm/values.yaml
• plugins/nemo-deployments/tests/unit/backends/docker/docker_helpers.py
• services/core/models/src/nmp/core/models/controllers/backends/deployments_plugin/config.py
• services/core/models/tests/unit/controllers/backends/deployments_plugin/test_compiler.py

[albcui]

I think for good measure, let's just add "XDG_CONFIG_HOME": _LORA_SIDECAR_XDG_HOME to services/core/models/src/nmp/core/models/controllers/backends/deployments_plugin/compiler.py as well?

[github-actions]

Suite	Lines Covered	Line Rate	Branch Rate
Unit Tests	31633/40251	78.6%	63.2%
Integration Tests	18379/38186	48.1%	20.8%

[soluwalana]

/nvskills-ci