16 releases (9 stable)

new 1.0.8 Jun 23, 2024
1.0.7 Jun 19, 2024
1.0.5 Apr 30, 2024
1.0.1 Mar 3, 2024
0.0.4 Dec 29, 2023

#122 in Database interfaces

Download history 172/week @ 2024-03-03 44/week @ 2024-03-10 1/week @ 2024-03-17 87/week @ 2024-03-31 194/week @ 2024-04-07 6/week @ 2024-04-14 125/week @ 2024-04-21 153/week @ 2024-04-28 3/week @ 2024-05-05 9/week @ 2024-05-19 113/week @ 2024-06-09 201/week @ 2024-06-16

316 downloads per month

MIT license

2.5K SLoC


Geni is a standalone migration tool designed to work in conjunction with your preferred ORM/toolkit/code. It allows multiple developers to collaborate without overriding database migrations. It can be used in a CD pipeline alongside your code to ensure your database stays up to date.

This project was heavily inspired by dbmate and was created because dbmate lacked support for LibSQL.

The application is developed using the Rust programming language and relies on the libsql-client-rs library for SQLite and LibSQL. Moreover, it makes use of SQLX to support Postgres, MariaDB, and MySQL databases. As this is written in rust is lighting fast, blazingly fast, tiny, ultra fast and memory safe

Geni in the news


  • Databases:
    • Postgres
    • MariaDB
    • MySQL
    • SQLite
    • LibSQL
  • Generating migrations using geni new **name**
  • Migrating using geni up
  • Rollback using geni down
  • Create database using geni create
  • Dropping database using geni drop
  • Timestamp based migrations
  • Running migrations in a transaction
  • Status command to see which migrations that is pending to be applied
  • Dump a schema.sql after each migration which can be used in version control
    • Dumping needs another binaries to work:
      • Postgres: pg_dump need to be installed(already installed in docker)
      • MySQL: mysqldump need to be installed(already installed in docker)
      • MariaDB: mariadb-dump need to be installed(already installed in docker)
      • Sqlite: Works without need for another binary. Uses SQL code get schema
      • LibSQL: Works without need for another binary. Uses SQL code get schema


  • Databases
    • ClickHouse



$ sudo curl -fsSL -o /usr/local/bin/geni https://github.com/emilpriver/geni/releases/latest/download/geni-linux-amd64
$ sudo chmod +x /usr/local/bin/geni


brew install geni




Run using PKGX

pkgx geni up

Nix flake

Run using nix

nix run github:emilpriver/geni -- up


cargo install geni


Docker images are published to GitHub Container Registry (ghcr.io/emilpriver/geni).

$ docker run --rm -it --network=host ghcr.io/emilpriver/geni:latest --help

there is also a slim docker image that don't have each database respective libraries(such as pg_dump).

Note: This image won't try to dump the database

$ docker run --rm -it --network=host ghcr.io/emilpriver/geni:latest-slim --help

If you wish to create or apply migrations, you will need to use Docker's bind mount feature to make your local working directory (pwd) available inside the geni container:

$ docker run --rm -it --network=host -v "$(pwd)/migrations:/migrations" ghcr.io/emilpriver/geni:latest new create_users_table`


geni new    # Generate a new migrations file
geni up     # Run any pending migration
geni down   # Rollback migrations, use --amount to speify how many migrations(default 1)
geni create # Create the database, only works for Postgres, MariaDB and MySQL. If you use SQLite will geni create the file before running migrations if the sqlite file don't exist. LibSQL should be create using respective interface.
geni drop   # Remove database
geni status # Print pending migrations
geni help   # Print help message

Environment variables

    • Specify where geni should look for migrations to run.
    • Default: ./migrations
    • The database url geni should use to make migrations
    • Examples:
      • Postgres:DATABASE_URL="postgres://postgres@"
      • MySQL: mysql://root:password@localhost:3307/app
      • MariaDB: mariadb://root:password@localhost:3307/app
      • Sqlite: sqlite://./database.sqlite
      • LibSQL: https://localhost:6000
        • The protocol for LibSQL is https.
        • For turso uses: This is something you can retrieve using Turso CLI or the website
    • Only if you use Turso and LibSQL and require token to authenticate. If not specified will Geni try to migrate without any auth
    • Time for geni to wait before trying to migrate. Useful if your database need some time to boot
    • Default: 30 seconds
    • Name of the schema migration file
    • Name of the table to run migrations to


Creating a new migration


DATABASE_URL="x" geni new hello_world

Will create 2 files(with path written in console). 1 file ending with .up.sql and 1 ending with .down.sql. .up.sql is for creating migrations and .down.sql is for rollbacking migrations. This means that .down.sql should include information that rollback the changed you added to .up.sql.


If I want to create table named Persons should I add this to .up.sql

    PersonID int

And the rollback migration should then be


in the generated .down.sql file as this code would revert the creation of the table Persons


Geni defaults to always run in transactions but if you want to prevent usage of transactions, add transaction: no as the first line of the migration file. Then Geni won't use transactions for the specific migration. This works for both up and down


-- transaction:no
CREATE TABLE table_2 (
  name VARCHAR(255) NOT NULL,

Running migration

Running migration can be done using

geni up

Rollback migrations

Rollbacking last added migrations can be done using

geni down

and If you want to rollback more then 1 can you add -a to the cli to specify how many

geni down -a 3

Running from CLI

DATABASE_URL="postgres://postgres@" geni up

Github Workflow

- uses: emilpriver/geni@main
    migrations_folder: "./migrations"
    wait_timeout: "30"
    migrations_table: "schema_migrations"
    database_url: "https://localhost:3000"
    database_token: "X"


  • migrations_folder(optional): The path to where your migrations exist.
    • Default: ./migrations
  • wait_timeout(optional): The time to wait before dropping the attempt to connect to the database
    • Default: 30
  • migrations_table(optional): The name of the migrations table
    • Default: schema_migrations
  • database_url(required): The url for accessing your database
  • database_token(optional): The token used to authenticate towards Turso. Only needed if you need to authenticate yourself
    • Default: ""

Running in CI/CD

In a CI/CD should the database_url come from a security store and be appended to the environment as the DATABASE_URL. If the DATABASE_URL is provided as a environment variable is the only command you need to run

geni up

to make migrations.

Running Geni as a library

Geni can be used as a library as well.

All exposed functions can be found in the library example folder

use geni;

async fn main() {
    // Migrate the database
        "sqlite://./test.db".to_string(), // Database URL
        None,                             // Database Token
        "migrations".to_string(),         // Migration Table
        "./migrations".to_string(),       // Migration Folder
        "schema.sql".to_string(),         // Schema File
        Some(30),                         // Wait timeout for the database to be ready
        false,                            // Dump Schema



~1M SLoC