Database migrations that stopped timing out
I moved TypeORM migrations off an API Gateway endpoint capped at 30 seconds onto a dedicated Lambda with a 15-minute window, run idempotently at three points of the release flow.
Outcome
- 30 s to 15 min migration time limit
- 3 automated run points
- 1 onboarding email, not 2
Stack
Ask me about this
The problem
Schema migrations ran through an API Gateway endpoint capped at 30 seconds. Any long change timed out halfway.
What I built
- A dedicated AWS Lambda for TypeORM migrations, with a 15-minute window, so long schema changes stopped timing out.
- It runs idempotently at 3 points: every staging deploy, the end of the nightly replication Step Function, and behind an approval gate in production.
Before API Gateway endpoint
Timed out
After Dedicated AWS Lambda
Done
Runs idempotently at 3 points
- 1 Every staging deploy
- 2 End of the nightly replication
- 3 Production, behind approval
Illustration with time sped up. The limits are the real ones: 30 seconds, and 15 minutes.
Product work
- Redesigned onboarding for law-firm admin users, replacing two conflicting verification and password emails with a single verify-and-set-password flow.
- Shipped platform features across React and Express, including cross-tab cache synchronization and bulk provider connection management.
Measured
Time limit for schema migrations
Rayna CorpMoved migrations off a capped API Gateway endpoint onto a dedicated Lambda, so long changes stopped timing out.
Onboarding emails for new admins
Rayna CorpReplaced separate verification and password emails for law-firm admins with a single verify-and-set-password flow.