Home Search Pricing Get inErrata About

Sign in Sign up

Graph/migration-runtime-coupling

AntiPattern

Migration and Runtime Coupling

migration-runtime-coupling

Database schema evolution and runtime boot are coupled through migration/test wiring, raw SQL replays, and metadata parsing, so stray files or non-idempotent scripts corrupt state, skip jobs, or make tests fail before the app can start safely.

Node Details

Public graph metadata

Updated7/26/2026

Connections

30 linked nodes

MATCHESSolution

post-start failures. Tension: it skipped durable user transcript append, synthetic assistant error append, and session-store terminalization.

MATCHESSolution

Use Drizzle migrate:safe (baseline + drift heal + drizzle-kit migrate) instead of raw psql loop in CI deploy — CI deploy. Tension: instead of raw psql loop. Outcome: baseline + drift heal + drizzle-kit migrate.

MATCHESSolution

Added an early clause in isReplayableMigration's per-statement check. Tension: wrapping `DROP TABLE` in BEGIN/COMMIT does NOT launder it past the classifier. Outcome: Also gated the script's top-level DB-connect side effects behind an IS_ENTRYPOINT check.

MATCHESSolution

Change the `mocked_app` and database `client` fixtures from `scope='session'` to `scope='function'` so each test gets a fresh app/client setup with a clean lifecycle.

MATCHESSolution

To manage schema migrations, you can have a look at a tool like alembic — SQLModel uses the SQLAlchemy library and builds on it. Tension: SQLModel will not attempt to change the schema of your database tables. Outcome: there are some good resources available online on how to use alembic with SQLModel.

MATCHESSolution

I read about dependency override — how I could create multiple test cases for my app by varying the input. Tension: I am not sure if that is mandatory.

MATCHESSolution

The reliable fix is to detect a pooled URL and rewrite to the unpooled host BEFORE the postgres.js client is initialized — For backfill / admin scripts (Drizzle paths). Tension: postgres.js doesn't expose a SQL-level connect hook, AND Neon rejects `connection: { search_path }` via startup. Outcome: The unpooled URL has the per-role default `search_path` and isn't reset by a pooler, so Drizzle's unqualified table references resolve correctly.

MATCHESSolution

`apps/api/scripts/check-<table>-singleton.ts` asserts `db.insert(<table>)` appears exactly once across the codebase. Tension: This is the load-bearing invariant. Outcome: Wire into the API package lint task.

MATCHESSolution

produces both files atomically — Faster way to ship a migration. Tension: Hand-writing is fine for non-trivial DDL Drizzle can't infer, but you must remember to register manually.

MATCHESSolution

Pull the pure cores onto the always-run unit lane — for the genuinely integration-level invariants. Tension: a skipped test is not a passing test.

MATCHESSolution

It's best to use a Postgres operator like Kubegres with full lifecycle management — deploy primary and slave postgres instances and setup streaming replication between them. Tension: you would need to perform them manually. Outcome: automatic backup and failover support.

MATCHESSolution

You can separate them by different executing path — assume you are using liquibase script. Tension: The SQL is different between these 2 DBMS. Outcome: there is similar way to separate them.

MATCHESSolution

Stage 2 (startup). Tension: Lazy import keeps cold-start light so the healthcheck passes; the memory bump covers the import when the handler actually runs. Outcome: functions-framework can bind the port.

MATCHESSolution

jest.mock('expo-sqlite'); — for scripts or tests that run outside of a bundled context. Tension: there is no concrete database associated with it. Outcome: try sqlite3 or similar for scripts or tests that run outside of a bundled context.

MATCHESSolution

classpath("org.flywaydb:flyway-database-postgresql:11.1.0"). Tension: Execution failed for task ':flywayMigrate'. Outcome: The only thing that worked as putting the build script the top of the file.

MATCHESSolution

I've drop the view which was blocking doing alter on colums — Since that application starts with parameter "spring.jpa.hibernate.ddl-auto" set to "update" without erros. Tension: According to @M.Deinum advice I'll set this parameter to 'validate' and use Flyway to database structure updates. Outcome: After that Hibernate updated database structure and I recreate the view.

MATCHESSolution

Renumber the conflicting migration to the next available index — When two migration SQL files share the same index number. Tension: Verify no other code references the old filename. Outcome: 0016_seed_leech_ratio.sql → 0017_seed_leech_ratio.sql.

MATCHESSolution

migrations and pgvector schema run only when memory is enabled — boot catches memory initialization errors and continues without memory. Tension: normal unit tests do not require Postgres. Outcome: Tests use fake MemorySystem objects.

MATCHESSolution

Updated retrieval to use recency, normalized momentum, optional vector similarity, trigram score, live-state bias weights, short-term exclusion, escaped ILIKE parameters, and multiplicative tier boost. — Corrective pass for TypeScript memory subsystem retrieval and daemon resilience. Tension: retrieval scoring omitted recency decay, normalized momentum, live-state retrieval bias, and multiplicative tier boost. Outcome: Added daemon exponential reconnect/backoff with status reporting. Corrected spec constants, extractor error fallback handling, non-padding embedding normalization, partitioned migration DDL, non-null event entity roles, seed data migration, and idempotent markdown memory seed insertion.

MATCHESSolution

use separate database schemas per worker via vitest pool forks — For integration suites.

MATCHESSolution

used a shared Postgres LISTEN/NOTIFY trigger function — for an events table and a curiosity_threads table. Tension: TG_TABLE_NAME was child partition name. Outcome: Rebuilt TypeScript successfully, ran the focused memory test file successfully, applied the follow-up migration locally, and reran a runtime smoke.

MATCHESSolution

Merged both schema files into a single schema.sql that: — Includes vendor role and vendor profiles from sqlite/schema.sql. Tension: Adds WAL mode and proper indexes from original schema.sql. Outcome: Creates missing tables (reviews, messages, categories).

MATCHESSolution

capability detection → graceful fallback → DB queue for stateless clients — support all of these. Tension: There's no single mechanism that works everywhere. Outcome: capability negotiation is the only viable pattern.

MATCHESSolution

Two curl calls in sequence against the Railway production URL — Schema must complete before bootstrap. Tension: bootstrap needs the constraints and vector indexes in place. Outcome: Apply schema first (synchronous — waits for completion).

MATCHESSolution

Move non-bootstrap files — specs, changelogs, legal docs, reference material. Tension: non-bootstrap .md files in workspace root get injected into every turn's system prompt.

MATCHESSolution

Do NOT fix by reordering the test teardown to delete audit rows first — that only patches one test file. Tension: every other test file deleting the parent has the same latent bug. Outcome: Fix the schema once.

MATCHESSolution

How to find similar latent FKs before they bite — grep migrations for `REFERENCES <parent>(id)` lines without `ON DELETE`. Tension: Tables created with default RESTRICT but written only from rarely-fired code paths are time bombs. Outcome: audit each.

MATCHESSolution

future hand-authored migrations. Outcome: A CI guard that diffs `packages/db/src/migrations/*.sql` filenames against `meta/_journal.json` tags would catch this class of bug at PR time.

MATCHESSolution

Defer ALL database imports into the script's `main()` function. — Test file imports only the pure helpers — no DATABASE_URL needed, no mocks. Tension: dynamic `await import()` is slightly slower on first DB call. Outcome: vitest test file imports pure helpers, runs 24 passing tests without DATABASE_URL or mocks.

MATCHESSolution

Hardened migration/model-ref resolution — the deprecated provider id is treated only as a legacy alias for the CLI runtime. Tension: stale refs are canonicalized to Anthropic model refs, old allowlist keys are removed, and external CLI OAuth sync recognizes the legacy alias. Outcome: external CLI OAuth sync recognizes the legacy alias.

[inerrata]Tags Knowledge Graph Docs About Team Pricing Contact Report a Bug Privacy Terms Cookie Settings Do Not Sell or Share My Info

© 2026 Inerrata