fix(api): surface shared-store (Redis/Upstash) liveness in /health [BUG-111]

[Morenikeoa]

Bug

UpstashStore is designed to fall back to an in-memory store on any Redis error, which is the right call-site behavior (a Redis outage shouldn't take the API down). But every fallback decision today is only visible via a scattered logger.warn() inside whichever method happened to fail. There's no operator-facing signal that the multi-replica abuse-control guarantee (rate limits, WS connection caps, auth bans — all documented at the top of shared-store.ts) has silently degraded to per-replica enforcement across the whole deployment. /health has no idea this subsystem exists.

Fix

Add ping(): Promise<boolean> to the SharedStore interface:

• InMemoryStore.ping() is trivially always true — it's the deliberately chosen single-replica backend, not a degraded state.
• UpstashStore.ping() issues a real Redis PING and returns false on any failure — deliberately not routed through the fallback-to-memory path the other methods use, since the whole point here is to report the real backend's reachability rather than mask it.

health.ts now calls getSharedStore().ping() and reports it as a new checks.sharedStore field, following the same pattern as the existing db/rpc/ws checks.

Test plan

• Added ping() liveness probe (BUG-111) block to tests/middleware/shared-store.test.ts: InMemoryStore always true; UpstashStore true on PONG, false on network failure, false on a non-PONG response.
• Added shared store health check (BUG-111) block to tests/routes/health.test.ts: checks.sharedStore flips false when ping() resolves false or throws, true when it resolves true; updated the existing "all checks fail" test to also mock ping() failing so it still exercises the true all-down case.
• Verified both are genuine regressions: reverted src/middleware/shared-store.ts + src/routes/health.ts via git stash, confirmed all 8 new/updated tests fail against the unfixed code (missing ping on the interface/check), restored the fix.
• Full suite passes (301/302 — the 1 pre-existing failure is unrelated to this change, in adl.test.ts).
• tsc --noEmit clean.

Summary by CodeRabbit

• New Features

  • Added a shared-store connectivity check to the health endpoint, improving visibility into service health.
  • Health responses now include shared-store status alongside existing checks.

• Bug Fixes

  • Health status now reflects shared-store failures as degraded or down instead of appearing fully healthy.
  • Redis-backed shared-store checks now fail safely when connectivity is lost or responses are unexpected.

• Tests

  • Expanded coverage for shared-store liveness and health endpoint behavior.

[vercel]

@Princessdada is attempting to deploy a commit to the Khubair Nasir's projects Team on Vercel.

A member of the Team first needs to authorize it.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 638e5f77-7734-4160-b71c-4c234bb7a291

📥 Commits

Reviewing files that changed from the base of the PR and between b2751f4 and d2c1a05.

📒 Files selected for processing (4)

• src/middleware/shared-store.ts
• src/routes/health.ts
• tests/middleware/shared-store.test.ts
• tests/routes/health.test.ts

---

📝 Walkthrough

Walkthrough

SharedStore now exposes a ping() liveness probe. /health calls that probe and includes shared-store health in its overall status. Tests cover in-memory and Upstash ping behavior plus the updated health-route outcomes.

Changes

Shared store ping contract

Layer / File(s)	Summary
Shared store ping contract src/middleware/shared-store.ts, tests/middleware/shared-store.test.ts	SharedStore adds ping(), InMemoryStore returns true, UpstashStore issues Redis PING, and tests cover success, non-PONG, and failure cases.
Health route shared-store check src/routes/health.ts, tests/routes/health.test.ts	/health calls getSharedStore().ping(), adds checks.sharedStore, and the route tests assert the new healthy and unhealthy outcomes.

Sequence Diagram

sequenceDiagram
  participant Client
  participant HealthRoute
  participant SharedStore
  Client->>HealthRoute: GET /health
  HealthRoute->>SharedStore: ping()
  alt ping succeeds
    SharedStore-->>HealthRoute: true
  else ping throws or fails
    SharedStore-->>HealthRoute: false
  end
  HealthRoute-->>Client: status with checks.sharedStore

Loading

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• dcccrypto/percolator-api#17: Also changes /health behavior in src/routes/health.ts and tests/routes/health.test.ts.
• dcccrypto/percolator-api#79: Also updates /health in src/routes/health.ts and related tests, including health evaluation behavior.

Poem

A tiny bunny hopped in light,
To ping the store by moonlit night.
If Redis snoozed, I’d still know true,
And /health would tell the view.
Thump thump — the checks are neat and bright.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly matches the main change: adding shared-store liveness reporting to /health.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}