Skip to content

fix: harden Trino registry-only routes and fixture readiness - #1173

Draft
benben wants to merge 5 commits into
mainfrom
fix/trino-registry-fixture-readiness
Draft

fix: harden Trino registry-only routes and fixture readiness#1173
benben wants to merge 5 commits into
mainfrom
fix/trino-registry-fixture-readiness

Conversation

@benben

@benben benben commented Sep 11, 2026

Copy link
Copy Markdown
Member

Summary

  • Runtime fix: return JSON 404 for unregistered /bundles and /bundles/ API paths instead of serving the admin SPA. Registered bundle authentication and ordinary UI routes are unchanged.
  • Wait for the exact expected cell registry after control-plane mode changes in the isolated Trino fixture.
  • Retry transient connection failures and stale Service endpoints with bounded requests. Fail if the registry stays wrong or unavailable.
  • Check the expected tenant password file with test -r on every ready legacy or registered worker before initial writes. Scope permissions to the existing isolated fixture namespaces.
  • Preserve all registered-query, ownership and legacy-compatibility assertions. No production configuration or provisioning-readiness changes; document that coordinator reconciliation does not guarantee worker Secret-volume convergence.

Validation

  • The previous live run passed legacy compatibility, initial placement, tenant isolation and green hydration. Its next request raced Service convergence after a control-plane restart.
  • Four behavioral regressions exercise the actual shell helper: connection refusal, stale endpoint, persistent wrong registry and unavailable API.
  • Clean runs exposed worker Secret-volume lag before both legacy and new-cell initial writes. Twenty-four worker-gate cases cover all-worker checks, three-worker convergence, missing files, stale readiness, cross-cell identity rejection, unavailable pods, denied access and redacted diagnostics. Writes are never retried.
  • The next live gate timed out safely before initial writes. In-harness comparison proved the pinned client's request-timeout flag disables its implicit in-cluster configuration. External five-second process bounds preserve working credential discovery; redacted stage diagnostics remain for future failures.
  • The shared gate now covers both legacy tenants and the registered tenant. The complete live registry-only proof remains pending.
  • The latest live run passed legacy compatibility, registered-cell writes, tenant isolation, green hydration and registered queries after registry-only startup. It exposed the SPA fallback returning HTML 200 for the absent legacy bundle. Composed-router regressions now cover this source correction.
  • just test-mw-fixtures and just lint pass. Independent adversarial review approved the bounded fix.
  • Live registry-only validation must pass before this follow-up is ready to merge.

@benben
benben requested a review from a team September 11, 2026 14:54
@github-actions

github-actions Bot commented Sep 11, 2026

Copy link
Copy Markdown

Test Impact Plan

Deterministic summary of how this PR changes tests, CI runners, and coverage-risk signals.

Summary

Area Added Changed Deleted
Test files 1 6 0
E2E/journey files 0 0 0
Workflow files 0 0 0

Signals

  • Test cases: +2 / -0
  • Assertions: +39 / -1
  • Skips or known failures added: 0
  • Workflow continue-on-error added: 0
  • Workflow path filters added: 0
  • Test commands removed from justfile: 0
  • E2E/journey retry lines added: 0

Coverage risk: neutral or increased

No coverage-reduction warnings detected.

@bill-ph bill-ph left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

Reviewed the bounded registry-convergence helper and its behavioral coverage. It retries transient API/refused/stale responses with a finite timeout, validates the exact expected cell set, and fails with an actionable message rather than proceeding on stale configuration. No runtime or deployment configuration changes are introduced, and I found no P0 blocker requiring a code change. The PR remains draft pending the live registry-only validation.\n\n— Robo Bill

@benben benben changed the title test: wait for Trino registry after control-plane rollouts fix: harden Trino registry-only routes and fixture readiness Sep 11, 2026
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