docs: planner 3-tier documentation restructure#5876
Conversation
|
Hi @athreesh! We're currently migrating our docs from Sphinx ( Could you please also apply your changes to the corresponding file(s) in
Note: The Thanks for helping keep both doc sets in sync during the migration! |
This is the first phase of a comprehensive documentation overhaul for
Dynamo, using the Planner component as the model migration.
## What's in this commit
### DGDR-as-Entrypoint UX Analysis (dgdr-ux-analysis.md)
- Analyzes the proposal to make DynamoGraphDeploymentRequest the primary
Kubernetes entrypoint
- Identifies 5 current UX pain points (two entrypoints, prerequisites
cliff, doc fragmentation, etc.)
- Recommends tiered entrypoint: DGDR for production, DGD for quick-start
- Proposes 5 specific fixes with effort/impact assessment
### AGENTS.md (root + planner)
- Root AGENTS.md: covers full repo (Rust runtime, Python components, Go
operator), build/test/run commands, architecture overview, key abstractions
- Planner AGENTS.md: component-specific guidance with entry flow, connector
pattern, backend mappings, modification patterns, error hierarchy
### Planner 3-Tier Documentation (model migration)
Applies the proposed documentation hierarchy to the Planner as a template
for all other components:
- Tier 1 (components/src/dynamo/planner/README.md): Updated redirect stub
with links to Tier 2 + Tier 3
- Tier 2 (docs/planner/):
- README.md: Overview, feature matrix, quick start, config reference
- planner_guide.md: Deployment, DGDR workflow, configuration, load
predictors, virtual connector, Prometheus setup, troubleshooting
- planner_examples.md: DGDR YAML examples (AIC, online, MoE, ConfigMap,
mocker), advanced patterns (warm-start, Kalman tuning, virtual connector)
- Tier 3 (docs/design_docs/planner_design.md): Architecture, scaling
algorithm with formulas, connector design, performance interpolation,
known limitations, future work
All content from the original 4 files (867 lines) is preserved and
redistributed. Original files are kept for backwards compatibility
during migration.
## Remaining action items for the full docs overhaul
### Phase 1: Remaining component migrations (same pattern as Planner)
- [ ] Frontend: create docs/frontend/{README,frontend_guide,frontend_examples}.md + design doc
- [ ] Router: create docs/router/{README,router_guide}.md + kv_routing_design.md
- [ ] KVBM: create docs/kvbm/{README,kvbm_guide}.md + kvbm_design.md
- [ ] Profiler: create docs/profiler/{README,profiler_guide,profiler_examples}.md
- [ ] Per-component AGENTS.md for frontend, router, kvbm, profiler
### Phase 2: Backend documentation
- [ ] Create docs/backends/README.md with backend comparison table + selection guide
- [ ] Normalize docs/backends/{vllm,sglang,trtllm}/ to consistent structure
- [ ] Add Tier 1 redirect stubs in components/src/dynamo/{vllm,sglang,trtllm}/README.md
### Phase 3: Cross-cutting features
- [ ] Create docs/features/ directory
- [ ] Move multimodal docs to docs/features/multimodal/ with per-backend pages
- [ ] Create docs/features/lora/ with vLLM-specific guide
- [ ] Create docs/features/speculative_decoding/ overview
### Phase 4: Kubernetes README restructure (from UX analysis)
- [ ] Restructure docs/kubernetes/README.md: quick-deploy (DGD) then production-deploy (DGDR)
- [ ] Create docs/kubernetes/dgdr-vs-dgd.md decision guide
- [ ] Add examples/quickstart/dgdr-quickstart.yaml minimal template
- [ ] Move observability/ under docs/kubernetes/ or elevate to docs/observability/
### Phase 5: Cleanup
- [ ] Delete planner_intro.rst after Sphinx toctree is updated
- [ ] Update all cross-references from old files to new structure
- [ ] Convert remaining .rst files to .md
- [ ] Delete docs/frontends/ (move kserve.md to kubernetes/)
- [ ] Verify Sphinx and Fern builds pass with new structure
### Phase 6: AGENTS.md rollout
- [ ] AGENTS.md for each backend (vllm, sglang, trtllm)
- [ ] AGENTS.md for deploy/operator/ (Go operator)
- [ ] AGENTS.md for lib/ crates (runtime, kv-router, memory)
- [ ] AGENTS.md for benchmarks/
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Signed-off-by: athreesh <anish.maddipoti@utexas.edu>
Keep only the planner 3-tier documentation restructure. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> Signed-off-by: athreesh <anish.maddipoti@utexas.edu>
- Create fern/pages/planner/planner-guide.md (from docs/planner/planner_guide.md) - Create fern/pages/planner/planner-examples.md (from docs/planner/planner_examples.md) - Create fern/pages/design-docs/planner-design.md (from docs/design_docs/planner_design.md) - Update fern/versions/next.yml with new pages in Planner and Design Docs sections - Remove stale AGENTS.md reference from components/src/dynamo/planner/README.md Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> Signed-off-by: athreesh <anish.maddipoti@utexas.edu>
10cd40a to
6b0f58d
Compare
WalkthroughDocumentation for the Planner autoscaling component is added across multiple directories. Changes include a README update for the component, new design documentation files, comprehensive user guides and examples for deployment and configuration, and navigation entries to support the documentation structure. Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~12 minutes Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Fix all issues with AI agents
In `@fern/pages/design-docs/planner-design.md`:
- Line 191: The bullet "30-second startup delay: Hardcoded wait for component
registration. Should be replaced with runtime readiness probing." contains a
sentence fragment; update the second sentence so it has a clear subject (e.g.,
"This should be replaced with runtime readiness probing" or "We should replace
it with runtime readiness probing") to make the bullet a complete, grammatical
statement referring to the hardcoded wait for component registration.
- Around line 61-64: Update the Markdown code fence that contains the correction
formulas so it includes a language tag (use "text") after the opening ```;
specifically change the fence surrounding the lines with prefill_correction =
actual_ttft / expected_ttft and decode_correction = actual_itl / expected_itl
to start with ```text instead of ```, leaving the content unchanged.
- Around line 16-44: Locate the ASCII architecture diagram fenced code block in
planner-design.md (the large "Planner Component" diagram) and change its opening
code fence from ``` to ```text so the fence includes the language tag; ensure
the closing fence remains ``` and no other diagram content is modified.
In `@fern/pages/planner/planner-examples.md`:
- Line 237: The link "Model Cache PVC" in planner-examples.md points to a
non-existent anchor (`#model-cache-pvc-advanced`) in
../benchmarks/sla-driven-profiling.md; either update the anchor in
planner-examples.md to an existing section anchor from sla-driven-profiling.md
or add a matching section heading ("Model Cache PVC" or "Model Cache PVC —
Advanced") with that anchor to sla-driven-profiling.md so the link resolves;
locate the link text in planner-examples.md and the section headings in
sla-driven-profiling.md to make the change.
🧹 Nitpick comments (2)
fern/pages/planner/planner-examples.md (1)
93-93: Consider simplifying the phrasing."Prior to 0.8.1" could be simplified to "Before 0.8.1" for easier reading.
docs/planner/planner_examples.md (1)
88-88: Consider simplifying the phrasing."Prior to 0.8.1" could be simplified to "Before 0.8.1" for easier reading.
Co-authored-by: Hongkuan Zhou <tedzhouhk@gmail.com> Signed-off-by: dagil-nvidia <dagil@nvidia.com>
Co-authored-by: Hongkuan Zhou <tedzhouhk@gmail.com> Signed-off-by: dagil-nvidia <dagil@nvidia.com>
Co-authored-by: Hongkuan Zhou <tedzhouhk@gmail.com> Signed-off-by: dagil-nvidia <dagil@nvidia.com>
Co-authored-by: Hongkuan Zhou <tedzhouhk@gmail.com> Signed-off-by: dagil-nvidia <dagil@nvidia.com>
Co-authored-by: Hongkuan Zhou <tedzhouhk@gmail.com> Signed-off-by: dagil-nvidia <dagil@nvidia.com>
- Replace ASCII diagram with aligned version and add text lang tag - Add text lang tag to correction formula code fence - Fix sentence fragment in Known Limitations - Fix broken anchor link in planner_examples.md - Remove fern/ planner files (handled separately) - Restore fern/versions/next.yml to main state Signed-off-by: Dan Gil <dagil@nvidia.com> Co-authored-by: Cursor <cursoragent@cursor.com>
|
/ok to test d399361 |
- Add README, Guide, Examples to planner_intro.rst toctree - Add planner_design.md to Design Docs toctree in index.rst - Fixes Build Documentation CI failure (toc.not_included warnings) Signed-off-by: Dan Gil <dagil@nvidia.com> Co-authored-by: Cursor <cursoragent@cursor.com>
|
/ok to test 84ad3dd |
Add SPDX-FileCopyrightText and SPDX-License-Identifier headers to three files from the Planner PR (#5876) that were missing them: - docs/design_docs/planner_design.md - docs/planner/planner_guide.md - docs/planner/planner_examples.md Signed-off-by: Dan Gil <dagil@nvidia.com> Co-authored-by: Cursor <cursoragent@cursor.com>
Signed-off-by: athreesh <anish.maddipoti@utexas.edu> Signed-off-by: dagil-nvidia <dagil@nvidia.com> Signed-off-by: Dan Gil <dagil@nvidia.com> Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com> Co-authored-by: dagil-nvidia <dagil@nvidia.com> Co-authored-by: Hongkuan Zhou <tedzhouhk@gmail.com> Co-authored-by: Cursor <cursoragent@cursor.com>
Summary
Applies a 3-tier documentation hierarchy to the Planner component as a model migration template for all other Dynamo components. Also mirrors the new docs into the
fern/directory for the Sphinx-to-Fern migration.Planner 3-Tier Documentation
Tier 1 — Component README (redirect stub)
components/src/dynamo/planner/README.md: Updated redirect stub with links to Tier 2 + Tier 3Tier 2 — User-facing docs (
docs/planner/+fern/pages/planner/)Tier 3 — Design docs (
docs/design_docs/+fern/pages/design-docs/)Fern navigation
fern/versions/next.ymlto include Planner Guide, Planner Examples, and Planner Design pagesAll content from the original planner docs (867 lines) is preserved and redistributed across the new structure.
Where should the reviewer start?
docs/planner/README.md— Tier 2 overviewdocs/planner/planner_guide.md— Tier 2 deployment guidefern/versions/next.yml— nav changesSummary by CodeRabbit