A migration helper written in Go. Use it in your existing Golang code or run commands via the CLI.
GoCode import github.com/mattes/migrate/migrate
CLI go get -u github.com/mattes/migrate
Features
- 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
.
- PostgreSQL
- Cassandra
- SQLite
- MySQL (experimental)
- Bash (planned)
Need another driver? Just implement the Driver interface and open a PR.
# install
go get github.com/mattes/migrate
# 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
See GoDoc here: http://godoc.org/github.com/mattes/migrate/migrate
import "github.com/mattes/migrate/migrate"
// 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.
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
002_xxx.up.sql
002_xxx.down.sql
...
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.