Skip to content

contracts(OMN-11913): add omnimarket 824 evidence - #1541

Merged
jonahgabriel merged 2 commits into
mainfrom
jonah/omn-11913-occ-evidence
May 24, 2026
Merged

jonahgabriel merged 2 commits into
mainfrom
jonah/omn-11913-occ-evidence

Conversation

@jonahgabriel

@jonahgabriel jonahgabriel commented May 24, 2026 •

Copy link
Copy Markdown
Contributor

Summary

Verification

  • uv run pytest src/omnimarket/nodes/node_antipattern_index_effect/tests/ -q
  • uv run mypy src/omnimarket/nodes/node_antipattern_index_effect --strict
  • OMNIBASE_INFRA_PATH=/Users/jonah/Code/omni_home/omnibase_infra pre-commit run --files ...
  • env -u LC_ALL -u LC_CTYPE pre-commit run --files contracts/OMN-11913.yaml drift/dod_receipts/OMN-11913/...

Evidence-Ticket: OMN-11913
Evidence-Source: OCC#1541

@coderabbitai

coderabbitai Bot commented May 24, 2026 •

Copy link
Copy Markdown

Warning

Review limit reached

@jonahgabriel, we couldn't start this review because you've used your available PR reviews for now.

Your plan currently allows 5 reviews/hour. Refill in 20 minutes and 30 seconds.

Your organization has run out of usage credits. Purchase more in the billing tab.

⌛ How to resolve this issue?

After more review capacity refills, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans have higher rate limits than trial, open-source, and free plans. In all cases, review capacity refills continuously over time.

Please see our FAQ for further information.

ℹ️ Review info
⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 6fd01fb8-6abb-42fe-8cfa-2cf3d45714df

📥 Commits

Reviewing files that changed from the base of the PR and between a9c82c5 and 925b17c.

📒 Files selected for processing (4)
  • contracts/OMN-11913.yaml
  • drift/dod_receipts/OMN-11913/dod-antipattern-index-local-validation/command.yaml
  • drift/dod_receipts/OMN-11913/dod-occ-pr/command.yaml
  • drift/dod_receipts/OMN-11913/dod-omnimarket-pr-824/command.yaml
✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Commit unit tests in branch jonah/omn-11913-occ-evidence

Comment @coderabbitai help to get the list of available commands and usage tips.

@jonahgabriel
jonahgabriel force-pushed the jonah/omn-11913-occ-evidence branch from 3c56d4a to 61621f3 Compare May 24, 2026 19:04
@jonahgabriel
jonahgabriel merged commit 5b74662 into main May 24, 2026
33 checks passed
@jonahgabriel
jonahgabriel deleted the jonah/omn-11913-occ-evidence branch May 24, 2026 19:35
jonahgabriel added a commit that referenced this pull request Sep 17, 2026
… taken with it (#10077)

* evidence(OMN-10856): the AC5 load-and-drain harness, and the live run taken with it

Companion for omninode_infra#1541.

The 2026-09-15 AC5 NOT MET note named the gap precisely:
`scripts/msk_consumer_group_health_check.py` "measures lag and can enforce
--require-zero-lag, but generates no load; it is a drain-verifier, not a load
generator", and "no raw producer was used and no live publish was attempted".
#1541 closes both, as a third mode of that same script rather than a second
script with a second credential path.

NOTHING HERE BINDS AC5, deliberately. AC5 is settled by measuring a running
system, and every check in this companion reads the repository tree or a
receipt. The OMN-18577 static-to-live proof-class rule refuses exactly that
pairing and is right to: a pytest proves the harness decides correctly, never
that lag drained on onex-dev. Ticking AC5 is the owner's call and belongs in
the ticket's Evidence citations, not in a binding minted here.

Two items. The first runs the harness's own decision logic against
protocol-shaped fakes -- six cases including the two that matter most, that a
drain with no publish would pass vacuously and that a topic no group consumes
must refuse at exit 2 rather than read zero lag as a drain. The second records
the live run, which CI cannot re-perform: the harness runs by `kubectl exec`
inside a runtime pod over that pod's own IAM-SASL identity, unreachable from a
hosted runner, so per OMN-14051 the probe ran out of band and its receipt is
what CI reads.

Live, twice, on 2026-09-17 against onex.evt.omnimarket.node-generation-completed.v1:
baseline 0, 10 published, peak 10, final 0, drained in 5.123s and 5.168s
against a stated 120s bound. Peak == published in both runs, which is the
positive control that the publish landed and was visible before it drained.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PacR7UupWUbZ2Hf8YnyXZV

* evidence(OMN-10856): self-bind to OCC companion PR #10077

The occ-preflight missing_occ_self_bind leg requires the PR carrying a contract
to bind itself, which needs the PR number and so could not land in the first
commit. Appending it moves contract_sha256, so both AC5 receipts are resynced;
their contract_entry_sha256 values are recomputed rather than assumed.

Verified with the same validator the gate runs: no missing receipts for this
ticket.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PacR7UupWUbZ2Hf8YnyXZV

* fix(OMN-10856): bind live drain receipt to product commit

* fix(OMN-10856): expand remaining receipt commit bindings

---------

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: jonahgabriel <jonah@omninode.ai>
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.

1 participant