Database migration handling in Golang

go get



Build Status GoDoc

A migration helper written in Go. Use it in your existing Golang code or run commands via the CLI.

GoCode   import
CLI      go get -u


  • Super easy to implement Driver interface.
  • Gracefully quit running migrations on ^C.
  • No magic search paths routines, no hard-coded config files.
  • CLI is build on top of the migrate package.

Available Drivers

Need another driver? Just implement the Driver interface and open a PR.

Usage from Terminal

# install
go get

# create new migration file in path
migrate -url driver://url -path ./migrations create migration_file_xyz

# apply all available migrations
migrate -url driver://url -path ./migrations up

# roll back all migrations
migrate -url driver://url -path ./migrations down

# roll back the most recently applied migration, then run it again.
migrate -url driver://url -path ./migrations redo

# run down and then up command
migrate -url driver://url -path ./migrations reset

# show the current migration version
migrate -url driver://url -path ./migrations version

# apply the next n migrations
migrate -url driver://url -path ./migrations migrate +1
migrate -url driver://url -path ./migrations migrate +2
migrate -url driver://url -path ./migrations migrate +n

# roll back the previous n migrations
migrate -url driver://url -path ./migrations migrate -1
migrate -url driver://url -path ./migrations migrate -2
migrate -url driver://url -path ./migrations migrate -n

# go to specific migration
migrate -url driver://url -path ./migrations goto 1
migrate -url driver://url -path ./migrations goto 10
migrate -url driver://url -path ./migrations goto v

Usage in Go

See GoDoc here:

import ""

// Import any required drivers so that they are registered and available
import _ ""

// use synchronous versions of migration functions ...
allErrors, ok := migrate.UpSync("driver://url", "./path")
if !ok {
  fmt.Println("Oh no ...")
  // do sth with allErrors slice

// use the asynchronous version of migration functions ...
pipe := migrate.NewPipe()
go migrate.Up(pipe, "driver://url", "./path")
// pipe is basically just a channel
// write your own channel listener. see writePipe() in main.go as an example.

Migration files

The format of migration files looks like this:

001_initial_plan_to_do_sth.up.sql     # up migration instructions
001_initial_plan_to_do_sth.down.sql   # down migration instructions

Why two files? This way you could still do sth like psql -f ./db/migrations/001_initial_plan_to_do_sth.up.sql and there is no need for any custom markup language to divide up and down migrations. Please note that the filename extension depends on the driver.