Task: Verify worker queue drain and health reporting

[KazanderDad]

Parent Goal: #80
Depends on: Scheduler configuration task
Blocks: Production smoke confidence

Objective

Verify worker queue drain and health reporting for the Supabase-side scheduled worker path.

Expected Surfaces

• Indexer touch/request queue
• Worker drain command/function
• Release-health RPC/workflow output
• Operator health/status views where available

Scope

• Queue or identify a safe worker request.
• Verify the scheduler/worker drains it.
• Confirm health reflects recent completion or failure state.
• Record evidence on Goal Goal: Prove Supabase-side worker scheduling #80.

Non-Goals

• Do not run unsafe mainnet value-moving operations.
• Do not expose raw secrets, wallet addresses, or service-role errors.

Guardrails

• Prefer Preview/dev proof before Production.
• Treat missing observability as a release blocker.

Acceptance Criteria

• Queue drain proof is captured or exact blocker is recorded.
• Health output shows safe queue status and recent run state where available.

Validation Plan

• Queue health query/RPC/workflow output.
• Worker run evidence.

Stop Condition

Stop after drain/health evidence is captured on Goal #80.

[KazanderDad]

Vetting Note

This issue is part of Sprint #78 and is vetted for the shared implementation branch/worktree below.

Implementation Worktree

• Repo: GreenPill-TO/Genero
• Base branch: origin/dev
• Worktree path: /Users/botmaster/src/greenpill-TO/Genero
• Worktree branch: codex/tcoin-production-launch-readiness
• Use this same worktree for related Sprint Sprint: TCOIN production launch readiness #78 issues: Goal: Align hosted release environments and remote Supabase readiness #74, Goal: Prove Preview sandbox smoke and Buy TCOIN sandbox #79, Goal: Prove Supabase-side worker scheduling #80, Goal: Prove Production smoke and live Buy TCOIN gate #81, and Tasks Task: Configure GitHub Environment secrets and vars for TCOIN release alignment #82-Task: Enforce live Buy TCOIN approval gate and disabled fallback #93.
• Worktree owner: Sprint Sprint: TCOIN production launch readiness #78 / sprint-orchestrator.
• Retain until: Sprint PR is merged and approved-cleanup completes.
• Cleanup owner: sprint-orchestrator via approved-cleanup.
• Do not implement the parent Sprint issue directly; implement unblocked Ready Task issues first, then validate their parent Goal.

Implementation Expectations

• Update code, docs, tests, and agent-context/session-log/ as relevant to the issue.
• Capture evidence directly on the relevant Goal issue, not in a separate evidence packet.
• Never print or persist secret values.
• Do not run linked Supabase writes without explicit just-in-time human approval.
• For UI/browser smoke, provide route/result notes and screenshots or comparable visual evidence when practical.

Issue Validator

After implementation, local validation, session-log update, and commit, the orchestrator must run issue-validator before moving high-risk Tasks or Goals to In Review.

Validator input should include: issue URL, parent issue URL, commit hash, worktree path, changed files summary, validation commands/results, smoke evidence, visual evidence where relevant, non-goals/guardrails, skipped checks, and residual risks.

Expected validator output: pass, pass-with-notes, or fail, with concise evidence and required fixes. The validator reports back to the orchestrator; the orchestrator moves Project status only after validation passes.

[KazanderDad]

Implementation evidence

Commit: 304d773 (fix: bound indexer queue drain worker (#89))

What changed:

• Added a bounded timeout path to drainIndexerTouchQueueOnce(...).
• scripts/indexer-touch-worker.ts now defaults INDEXER_TOUCH_WORKER_TIMEOUT_MS to 240000 and exits non-zero on failure.
• If a claimed worker times out, the worker marks indexer.run_control as error and marks the claimed touch request as failed, preventing an indefinite running health state.
• Updated the release runbook with timeout configuration and repeated-timeout handling guidance.
• Added focused unit coverage for scoped claims and timeout failure handling.

Validation:

• pnpm exec vitest run services/indexer/src/touchQueue.test.ts
• pnpm exec tsc --noEmit --pretty false
• pnpm lint
• git diff --check
• pnpm ops:wallet:preflight:supabase-remote
• pnpm exec tsx scripts/run-with-env-profile.ts .env.local-supabase-remote -- pnpm ops:torontocoin

Read-only remote health evidence:

• pnpm ops:wallet:preflight:supabase-remote returned blockers: [].
• Cleanup cron latest status is succeeded at 2026-07-20T06:15:00Z.
• Indexer queue: pendingRequestCount=0, blocked=false, stale=false.
• pnpm ops:torontocoin returned releaseBlockers: [] and Indexer queue pending: 0, Indexer queue blocked: false.

Boundary check:

• No linked Supabase writes were run.
• No queue request was created or drained against Preview/Production because this turn has no just-in-time approval for linked writes.
• Exact remaining blocker: a real hosted queue-drain proof requires approval to create or identify a safe Preview queue request and run the scoped worker once.

[KazanderDad]

Issue Validator result

Result: pass-with-notes

Target issue: #89
Commit reviewed: 304d773

Evidence reviewed:

• Implementation evidence comment on Task: Verify worker queue drain and health reporting #89.
• Goal Goal: Prove Supabase-side worker scheduling #80 queue-drain progress comment.
• Commit diff for bounded worker timeout, CLI timeout handling, runbook docs, and tests.
• Read-only remote health outputs from wallet preflight and TorontoCoin ops.

Validation rerun:

• pnpm exec vitest run services/indexer/src/touchQueue.test.ts passed.
• git diff --check passed.
• Previously recorded implementation validation also included pnpm exec tsc --noEmit --pretty false and pnpm lint, both passing.

Boundary check:

• No linked Supabase writes were run.
• No unsafe mainnet/value-moving operations were run.
• No raw secrets or service-role errors were exposed.

Notes:

• The task's health-reporting requirement is satisfied for read-only Preview health: queue pending 0, blocked false, stale false, and remote release blockers empty.
• This is pass-with-notes because a live queue-drain proof against Preview still requires just-in-time approval to create or identify a safe queued request and run the scoped worker once.

Ready for Project status: In Review.