[cristim]

Summary

Two migration-safety defects found by adversarial review of committed origin/main, both in internal/database/postgres/migrations/migrate.go and its tests:

Defect 1 (HIGH): maybeAutoHealDirty silently marks a rolled-back migration as applied

golang-migrate records version=N, dirty=true BEFORE running N's SQL. Every migration in this repo is transactional (TestMigrationsAreTransactional), so an interrupted migration (Lambda timeout, connection drop) rolls back completely -- N's effects are ABSENT -- but the dirty flag persists.

The old Force(N)+Up() auto-heal treated the rolled-back migration as fully applied and continued at N+1, permanently skipping N's DDL while schema_migrations reported success. This is the 000074 divergence class: 42703 "column does not exist" errors at runtime on columns the version record claimed were present.

There is also a narrow race where N commits but the dirty=false update fails (N's effects ARE present). These two cases are indistinguishable without per-migration effect probes, so the conservative fix is:

• maybeAutoHealDirty now returns a loud error with recovery instructions when dirty is detected, rather than silently clearing the flag.
• The app still fail-opens (ensureDB swallows the error) and the CloudWatch alarm fires; no crash-loop.
• Operators use CUDLY_FORCE_MIGRATION_VERSION=N (if effects confirmed) or N-1 (if not) to recover.
• CUDLY_MIGRATION_AUTOHEAL=false still bypasses the check (surfaces the generic Up() dirty error instead).

Defect 2 (MED): TestMigrations_FullStackIdempotent falsely claimed to guard per-migration idempotency

The test ran RunMigrations twice, but the second call returns ErrNoChange from golang-migrate without executing any SQL -- no migration body is actually re-run. The docstring incorrectly claimed "this is the invariant the opt-in CUDLY_MIGRATION_AUTOHEAL path relies on" and that it "guards the whole directory."

Fix: corrected the docstring to accurately describe the test (verifies a no-op second call on a fully-migrated DB) and removed the false auto-heal dependency claim. The conservative auto-heal no longer relies on per-migration idempotency.

Changes

• migrate.go: Replace Force(N)+continue in maybeAutoHealDirty with a loud error containing structured recovery instructions.
• migrate_autoheal_test.go: Update the two "self-heals" test cases to expect errors and verify the dirty flag is left intact. Add regression test dirty-at-N with N-effects-absent (rolls back last migration, injects dirty=true at head, asserts error + dirty flag preserved).
• migrate_idempotency_test.go: Correct the false docstring; remove the unverified "guards the whole directory" claim.

Test plan

• go build ./internal/database/... passes
• go vet ./internal/database/... passes
• go test ./internal/database/... (non-integration): 181 passed
• All pre-commit hooks pass (gofmt, go vet, cyclomatic complexity, gosec)
• Integration tests (require Docker): go test -tags integration ./internal/database/postgres/migrations/... -- CI will run these
• The three updated auto-heal test cases verify the new fail-loud behavior
• The new regression test dirty-at-N with N-effects-absent verifies Force(N) is NOT called on a rolled-back migration

[cristim]

@coderabbitai review

[coderabbitai]

Warning

Review limit reached

You’ve reached a temporary PR review limit under our Fair Usage Limits Policy.

Your recent review volume is higher than typical usage, so adaptive limits are currently applied.

Next review available in: 57 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

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

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 81fd0729-5018-4cc1-ad69-77b68de1005f

📥 Commits

Reviewing files that changed from the base of the PR and between 7465a46 and 5c6bf44.

📒 Files selected for processing (3)

• internal/database/postgres/migrations/migrate.go
• internal/database/postgres/migrations/migrate_autoheal_test.go
• internal/database/postgres/migrations/migrate_idempotency_test.go

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/migration-autoheal-safety

---

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

[coderabbitai]

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[cristim]

@coderabbitai full review

[coderabbitai]

✅ Action performed

Full review finished.

---

You're currently rate limited under our Fair Usage Limits Policy. Your recent PR review activity is in the 95th percentile or higher among CodeRabbit users, so adaptive limits apply. Your next review will be available in 57 minutes.