docs(miner): document and load-test shared store concurrency model

[andriypolanski]

Summary

• The miner's local-store concurrency story was written for two processes on one SQLite file; after Migrate AMS's local-store layer (node:sqlite) to the shared pg-adapter/SqliteDriver seam #7175's shared SqliteDriver seam that model needed an explicit shared-service restatement (Re-evaluate the local concurrency model for a shared-service context #4942).
• This adds the maintainer concurrency-model doc plus a correctness load suite (no lost updates / no double-claim under the patterns the model claims are safe).

Closes #4942

Change

• Doc packages/loopover-miner/docs/ams-shared-store-concurrency-model.md: what SQLite BEGIN IMMEDIATE / busy_timeout, miner createD1Adapter.batch, and ORB createPgAdapter (READ COMMITTED) guarantee — and what they do not (SERIALIZABLE, installation admission, two loops on one dir, miner-already-on-Postgres). Contrasts installation-concurrency-admission / mapWithConcurrency. Notes control-plane one-DB-per-tenant isolation.
• Links from operations-runbook.md and ams-storage-abstraction-research.md.
• Load / race tests test/unit/miner-shared-store-concurrency.test.ts + claim-within-cap-child.mjs:
  • Cross-process claimIssueWithinCap (cap=1, 8 distinct issues → exactly one active row)
  • 32× concurrent UPDATE … SET n = n + 1 via createD1Adapter.batch → exact count
  • Naive interleaved RMW schedule pins the documented lost-update non-guarantee
  • Optional Postgres suite (PG_TEST_URL) for the same atomic increment via createPgAdapter

Validation

• git diff --check clean.
• npx vitest run test/unit/miner-shared-store-concurrency.test.ts test/unit/miner-operations-runbook.test.ts — green (PG case skipped without PG_TEST_URL).
• Docs / fixtures / unit tests only — no src/** or miner lib/** executable changes; Codecov patch has no graded diff lines.

UI Evidence

N/A — documentation + concurrency correctness tests only.

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[loopover-orb]

Caution

🛑 LoopOver review result - fixes required

_{Review updated: 2026-07-22 06:59:15 UTC}

6 files · 1 AI reviewer · no blockers · CI failing · blocked

🛑 Suggested Action - Fix Blockers

Review summary
This PR adds a maintainer doc (`ams-shared-store-concurrency-model.md`) restating SQLite/Postgres concurrency guarantees for the shared-service model, links it from the runbook and storage-abstraction research doc, and backs it with a cross-process load-test suite (claimIssueWithinCap race, SqliteDriver/createD1Adapter atomic increments, an optional Postgres suite, and a deliberately-sequential 'lost update' demonstration). The diff is scoped correctly to docs+tests with no src/lib changes, closes #4942 as required, and the doc content matches what the new assertions check for (BEGIN IMMEDIATE, READ COMMITTED, SqliteDriver, installation-concurrency-admission, map-with-concurrency.ts substring). The most notable wrinkle is that the 'concurrent' SqliteDriver increment test drives `Promise.all` over calls into a synchronous driver, so it can't produce genuine interleaving in Node's single-threaded model — it mainly re-verifies sequential additive correctness rather than a real race, unlike the cross-process child-process test which does.

Nits — 5 non-blocking

• test/unit/miner-shared-store-concurrency.test.ts's 'N concurrent atomic UPDATE' SqliteDriver test wraps synchronous `adapter.batch()` calls in `Promise.all`, which cannot produce genuine thread interleaving in Node — it's effectively sequential execution dressed as concurrency, unlike the real cross-process `claimIssueWithinCap` test above it; consider noting this distinction in the test description or doc's 'Load / race verification' section so readers don't over-read it as a true race test.
• test/unit/miner-shared-store-concurrency.test.ts unconditionally imports `pg` and `../../src/selfhost/pg-adapter` at module scope even though the Postgres suite is skipped without `PG_TEST_URL` — confirm `pg` is already a resolvable devDependency for this workspace so the whole file (including the always-run SQLite tests) doesn't fail to load in environments without it.
• The two failing 'validate-tests' CI checks and 'validate' check are listed with no detail provided, so their cause could not be verified from what's given here — worth checking the actual CI logs before merge since the PR description claims a green local run.
• Consider asserting the doc's `mapWithConcurrency` guarantee text with a more literal match (currently the test's `map-with-concurrency` substring is satisfied only by the file path `map-with-concurrency.ts`, not the camelCase symbol name used in prose) so the assertion tracks the intended prose claim rather than an incidental file-path substring.
• In the 'naive app-level RMW' test, a short comment noting this is a deterministic single-threaded demonstration (not a true race) rather than empirical timing-based interleaving would make the non-guarantee documentation clearer for future readers.

CI checks failing

• validate
• validate-tests (1)
• validate-tests (3)
• validate-tests (2)

Decision drivers

• ✅ Code review — No blockers (1 reviewer)
• ✅ Gate result — Passing (No configured blocker found.)

Context & advisory signals — never blocks the verdict

Signal	Result	Evidence
Linked issue	✅ Linked	#4942
Related work	✅ No active overlap found	No same-issue or scoped active PR overlap found.
Change scope	✅ 20/20	Low review scope from cached public metadata (1 linked issue).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 199 registered-repo PR(s), 128 merged, 22 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	andriypolanski; Gittensor profile; 199 PR(s), 22 issue(s).
Improvement	ℹ️ Insufficient signal	risk: clean · value: insufficient-signal · LLM: moderate

Linked issue satisfaction

Addressed
The PR adds an explicit concurrency-model doc distinguishing SQLite/D1-adapter and Postgres pg-adapter guarantees (and non-guarantees) versus the old single-file assumptions, and includes cross-process/concurrent load tests (claim cap, atomic increments, a naive RMW lost-update demonstration) verifying no lost updates under real concurrent access, matching both deliverables in the issue.

Review context

• Author: andriypolanski
• Role context: outside_contributor
• Public audience mode: oss maintainer
• Lane context: Repository is configured for direct PR review.
• Public profile languages: Python, TypeScript, JavaScript, Rust, Cuda, Kotlin, MDX, Scala
• Official Gittensor activity: 199 PR(s), 22 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Keep the PR focused and include validation evidence before maintainer review.

Signal definitions

• Related work = same linked issue, overlapping active PRs, or title/path similarity.
• Change scope = cached public metadata such as size labels, draft state, and review-burden hints.
• Validation posture = whether the PR provides enough public validation/test evidence for maintainer review.
• Contributor workload = public contributor activity and cleanup pressure, not a repo-wide quality failure.
• Contributor context = public GitHub/Gittensor identity context; non-Gittensor status is not a blocker.

🧪 Chat with LoopOver

Ask LoopOver a question about this PR directly in a comment — grounded only in the same cached, public-safe facts shown above, never a new claim.

• @loopover ask <question> answers contribution-quality Q&A with source citations and freshness.
• @loopover chat <question> answers in natural prose from cached decision-pack facts via local inference (maintainer/collaborator; read-only).
• A plain-language @loopover mention with a real question is routed to the closest matching read-only command automatically — no exact syntax required.

Full command reference: https://loopover.ai/docs/loopover-commands

🧪 Experimental — new and may change.

_{🟩 Safe / merged · 🟦 Advisory · 🟨 Held for review · 🟥 Blocked / closed}

---

💰 Earn for open-source contributions like this. Gittensor lets GitHub contributors earn for the work they already do — register to start earning →.

Checked by LoopOver, a quiet PR intelligence layer for OSS maintainers.

• Re-run LoopOver review

[loopover-orb]

LoopOver is closing this pull request on the maintainer's behalf (CI is failing (validate, validate-tests (1), validate-tests (3), validate-tests (2))). This is an automated maintenance action — to pursue this change, please open a new pull request with the issues resolved. Closed PRs may be analyzed later to improve review accuracy, but they are not automatically reopened or re-reviewed.