Skip to content

workflows.md

Latest commit

 

History

History
51 lines (41 loc) · 1.96 KB

workflows.md

File metadata and controls

51 lines (41 loc) · 1.96 KB

Workflows

PGV is built to handle the messiest parts of local database development. Here are common scenarios where PGV shines.

Scenario 1: The "Destructive Migration" Test

You are about to run a massive data migration that deletes columns, transforms data, and takes 5 minutes to run. You want to test it locally, but if it fails, you don't want to spend 20 minutes running pg_restore to fix your local DB.

  1. Save your state: 
    pgv checkpoint "before massive migration"
  2. Run your migration (using Prisma, Drizzle, Flyway, etc).
  3. Did it fail? Roll back instantly: 
    pgv stop main
pgv restore "before massive migration"
pgv start main
    Your database is back online exactly as it was, in 2 seconds.

Scenario 2: Parallel Feature Development

You are reviewing a PR for "Feature A" which introduces a new orders table. You are actively coding "Feature B" which introduces a subscriptions table. You need to switch between them without the schemas clashing.

  1. Create branches from your clean main snapshot: 
    pgv branch snap_clean feature-a
pgv branch snap_clean feature-b
  2. Start both branches: 
    pgv start feature-a  # Runs on port 5541
pgv start feature-b  # Runs on port 5542
  3. Switch contexts effortlessly: You can point your application to 5541 to test the PR, and point it to 5542 to continue your own work. The databases are physically isolated.

Scenario 3: Seed Data "Save States"

Generating massive amounts of seed data locally takes a long time.

  1. Write a script to generate 10GB of fake data. Run it on your main branch.
  2. Save this state: 
    pgv checkpoint "10GB seed data"
  3. Now, whenever you need a fresh database full of data to test a new query, you can instantly branch off that snapshot without having to run the 10-minute seed script ever again: 
    pgv branch "10GB seed data" query-test-env