Skip to content

fix(paseo-omp): raise catalog budgets for large OMP registries - #276

Open
omercnet wants to merge 1 commit into
mainfrom
fix/paseo-omp-catalog-budgets
Open

omercnet wants to merge 1 commit into
mainfrom
fix/paseo-omp-catalog-budgets

Conversation

@omercnet

@omercnet omercnet commented Oct 4, 2026

Copy link
Copy Markdown
Owner

Why

A standard OMP 18.6 registry (962 models) serializes to 16385 nodes. That exceeded the get_available_models node budget (16384), so discovery failed with rpc-response-limit and left the provider in error. Separately, model selection kept only the first 256 models and silently dropped every provider after that.

Change

  • get_available_models response node budget 16384 -> 131072. A full 4096-model catalog is about 70k nodes (about 17 per model), so a smaller bound would still fail near the cap.
  • OMP_MAX_CATALOG_MODELS 256 -> 4096, matching the existing MAX_MODEL_CATALOG_ITEMS.
  • Scale the 257-model boundary tests to 4097 and the oversize fixture in omp-rpc-protocol.test.ts so it still exceeds the node budget.
  • Add a 1500-model regression test. It fails with the old 16384 budget and passes now.

Verification

  • npm run typecheck clean.
  • bun x biome check . clean.
  • bun x vitest run: 849 passed, 7 skipped, 0 failed.
  • Regression test confirmed red against the old limit.

Credit: the original fix is from @masrurimz (Zahid). This PR adds the budget alignment and tests.

Co-authored-by: Zahid masruri03@gmail.com

A standard OMP 18.6 registry (962 models) serializes to 16385 nodes and
exceeded the get_available_models node budget, so discovery failed with
rpc-response-limit and left the provider in error.

- get_available_models node budget 16384 -> 131072, so a full 4096-model
  catalog fits with headroom (about 17 nodes per model today)
- OMP_MAX_CATALOG_MODELS 256 -> 4096 (matches MAX_MODEL_CATALOG_ITEMS);
  selection previously dropped every provider past the first 256 models
- scale the 257-model boundary tests to 4097 and the oversize fixture so
  it still exceeds the node budget
- add a 1500-model regression test that fails on the old budget

Co-authored-by: Zahid <masruri03@gmail.com>
Copilot AI balanced review requested due to automatic review settings October 4, 2026 16:43

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants