Skip to main content

Migrations

How migrations work

Once you get into production you'll need to synchronize model changes into the database. Typically, it is unsafe to use synchronize: true for schema synchronization on production once you get data in your database. Here is where migrations come to help.

A migration is just a single file with sql queries to update a database schema and apply new changes to an existing database.

Let's say you already have a database and a post entity:

import { Entity, Column, PrimaryGeneratedColumn } from "typeorm"

@Entity()
export class Post {
@PrimaryGeneratedColumn()
id: number

@Column()
title: string

@Column()
text: string
}

And your entity worked in production for months without any changes. You have thousands of posts in your database.

Now you need to make a new release and rename title to name. What would you do?

You need to create a new migration with the following SQL query (postgres dialect):

ALTER TABLE "post" ALTER COLUMN "title" RENAME TO "name";

Once you run this SQL query your database schema is ready to work with your new codebase. TypeORM provides a place where you can write such sql queries and run them when needed. This place is called "migrations".

Creating a new migration

Pre-requisites: Installing CLI

Before creating a new migration you need to setup your data source options properly:

{
type: "mysql",
host: "localhost",
port: 3306,
username: "test",
password: "test",
database: "test",
entities: [/*...*/],
migrations: [/*...*/],
migrationsTableName: "custom_migration_table",
}

Here we setup two options:

  • "migrationsTableName": "migrations" - Specify this option only if you need migration table name to be different from "migrations".
  • "migrations": [/*...*/] - list of migrations need to be loaded by TypeORM

Once you setup connection options you can create a new migration using CLI:

typeorm migration:create ./path-to-migrations-dir/PostRefactoring

Here, PostRefactoring is the name of the migration - you can specify any name you want. After you run the command you can see a new file generated in the "migration" directory named {TIMESTAMP}-PostRefactoring.ts where {TIMESTAMP} is the current timestamp when the migration was generated. Now you can open the file and add your migration sql queries there.

You should see the following content inside your migration:

import { MigrationInterface, QueryRunner } from "typeorm"

export class PostRefactoringTIMESTAMP implements MigrationInterface {
async up(queryRunner: QueryRunner): Promise<void> {}

async down(queryRunner: QueryRunner): Promise<void> {}
}

There are two methods you must fill with your migration code: up and down. up has to contain the code you need to perform the migration. down has to revert whatever up changed. down method is used to revert the last migration.

Inside both up and down you have a QueryRunner object. All database operations are executed using this object. Learn more about query runner.

Let's see what the migration looks like with our Post changes:

import { MigrationInterface, QueryRunner } from "typeorm"

export class PostRefactoringTIMESTAMP implements MigrationInterface {
async up(queryRunner: QueryRunner): Promise<void> {
await queryRunner.query(
`ALTER TABLE "post" RENAME COLUMN "title" TO "name"`,
)
}

async down(queryRunner: QueryRunner): Promise<void> {
await queryRunner.query(
`ALTER TABLE "post" RENAME COLUMN "name" TO "title"`,
) // reverts things made in "up" method
}
}

Running and reverting migrations

Once you have a migration to run on production, you can run them using a CLI command:

typeorm migration:run -- -d path-to-datasource-config

typeorm migration:create and typeorm migration:generate will create .ts files, unless you use the o flag (see more in Generating migrations). The migration:run and migration:revert commands only work on .js files. Thus the typescript files need to be compiled before running the commands. Alternatively you can use ts-node in conjunction with typeorm to run .ts migration files.

Example with ts-node:

npx typeorm-ts-node-commonjs migration:run -- -d path-to-datasource-config

Example with ts-node in ESM projects:

npx typeorm-ts-node-esm migration:run -- -d path-to-datasource-config
npx typeorm-ts-node-esm migration:generate ./src/migrations/update-post-table -d ./src/data-source.ts

This command will execute all pending migrations and run them in a sequence ordered by their timestamps. This means all sql queries written in the up methods of your created migrations will be executed. That's all! Now you have your database schema up-to-date.

If for some reason you want to revert the changes, you can run:

typeorm migration:revert -- -d path-to-datasource-config

This command will execute down in the latest executed migration. If you need to revert multiple migrations you must call this command multiple times.

Faking Migrations and Rollbacks

You can also fake run a migration using the --fake flag (-f for short). This will add the migration to the migrations table without running it. This is useful for migrations created after manual changes have already been made to the database or when migrations have been run externally (e.g. by another tool or application), and you still would like to keep a consistent migration history.

typeorm migration:run --fake

This is also possible with rollbacks.

typeorm migration:revert --fake

Transaction modes

By default, TypeORM will run all your migrations within a single wrapping transaction. This corresponds to the --transaction all flag. If you require more fine grained transaction control, you can use the --transaction each flag to wrap every migration individually, or the --transaction none flag to opt out of wrapping the migrations in transactions altogether.

In addition to these flags, you can also override the transaction behavior on a per-migration basis by setting the transaction property on the MigrationInterface to true or false. This only works in the each or none transaction mode.

import { MigrationInterface, QueryRunner } from "typeorm"

export class AddIndexTIMESTAMP implements MigrationInterface {
transaction = false

async up(queryRunner: QueryRunner): Promise<void> {
await queryRunner.query(
`CREATE INDEX CONCURRENTLY post_names_idx ON post(name)`
)
}

async down(queryRunner: QueryRunner): Promise<void> {
await queryRunner.query(
`DROP INDEX CONCURRENTLY post_names_idx`,
)
}
}

Generating migrations

TypeORM is able to automatically generate migration files with schema changes you made.

Let's say you have a Post entity with a title column, and you have changed the name title to name. You can run following command:

typeorm migration:generate PostRefactoring -d path-to-datasource-config

If you encounter any error, it require you have the path to migration name and data source. You can try this option

typeorm migration:generate -d <path/to/datasource> path/to/migrations/<migration-name>

And it will generate a new migration called {TIMESTAMP}-PostRefactoring.ts with the following content:

import { MigrationInterface, QueryRunner } from "typeorm"

export class PostRefactoringTIMESTAMP implements MigrationInterface {
async up(queryRunner: QueryRunner): Promise<void> {
await queryRunner.query(
`ALTER TABLE "post" ALTER COLUMN "title" RENAME TO "name"`,
)
}

async down(queryRunner: QueryRunner): Promise<void> {
await queryRunner.query(
`ALTER TABLE "post" ALTER COLUMN "name" RENAME TO "title"`,
)
}
}

Alternatively you can also output your migrations as Javascript files using the o (alias for --outputJs) flag. This is useful for Javascript only projects in which TypeScript additional packages are not installed. This command, will generate a new migration file {TIMESTAMP}-PostRefactoring.js with the following content:

const { MigrationInterface, QueryRunner } = require("typeorm")

module.exports = class PostRefactoringTIMESTAMP {
async up(queryRunner) {
await queryRunner.query(
`ALTER TABLE "post" ALTER COLUMN "title" RENAME TO "name"`,
)
}

async down(queryRunner) {
await queryRunner.query(
`ALTER TABLE "post" ALTER COLUMN "name" RENAME TO "title"`,
)
}
}

See, you don't need to write the queries on your own. The rule of thumb for generating migrations is that you generate them after each change you made to your models. To apply multi-line formatting to your generated migration queries, use the p (alias for --pretty) flag.

DataSource option

If you need to run/revert/generate/show your migrations use the -d (alias for --dataSource) and pass the path to the file where your DataSource instance is defined as an argument

typeorm -d <your-data-source-path> migration:{run|revert}

Timestamp option

If you need to specify a timestamp for the migration name, use the -t (alias for --timestamp) and pass the timestamp (should be a non-negative number)

typeorm -t <specific-timestamp> migration:{create|generate}

You can get a timestamp from:

Date.now()
/* OR */ new Date().getTime()

Using migration API to write migrations

In order to use an API to change a database schema you can use QueryRunner.

Example:

import {
MigrationInterface,
QueryRunner,
Table,
TableIndex,
TableColumn,
TableForeignKey,
} from "typeorm"

export class QuestionRefactoringTIMESTAMP implements MigrationInterface {
async up(queryRunner: QueryRunner): Promise<void> {
await queryRunner.createTable(
new Table({
name: "question",
columns: [
{
name: "id",
type: "int",
isPrimary: true,
},
{
name: "name",
type: "varchar",
},
],
}),
true,
)

await queryRunner.createIndex(
"question",
new TableIndex({
name: "IDX_QUESTION_NAME",
columnNames: ["name"],
}),
)

await queryRunner.createTable(
new Table({
name: "answer",
columns: [
{
name: "id",
type: "int",
isPrimary: true,
},
{
name: "name",
type: "varchar",
},
{
name: "created_at",
type: "timestamp",
default: "now()",
},
],
}),
true,
)

await queryRunner.addColumn(
"answer",
new TableColumn({
name: "questionId",
type: "int",
}),
)

await queryRunner.createForeignKey(
"answer",
new TableForeignKey({
columnNames: ["questionId"],
referencedColumnNames: ["id"],
referencedTableName: "question",
onDelete: "CASCADE",
}),
)
}

async down(queryRunner: QueryRunner): Promise<void> {
const table = await queryRunner.getTable("answer")
const foreignKey = table.foreignKeys.find(
(fk) => fk.columnNames.indexOf("questionId") !== -1,
)
await queryRunner.dropForeignKey("answer", foreignKey)
await queryRunner.dropColumn("answer", "questionId")
await queryRunner.dropTable("answer")
await queryRunner.dropIndex("question", "IDX_QUESTION_NAME")
await queryRunner.dropTable("question")
}
}

getDatabases(): Promise<string[]>

Returns all available database names including system databases.


getSchemas(database?: string): Promise<string[]>
  • database - If database parameter specified, returns schemas of that database

Returns all available schema names including system schemas. Useful for SQLServer and Postgres only.


getTable(tableName: string): Promise<Table|undefined>
  • tableName - name of a table to be loaded

Loads a table by a given name from the database.


getTables(tableNames: string[]): Promise<Table[]>
  • tableNames - name of a tables to be loaded

Loads a tables by a given names from the database.


hasDatabase(database: string): Promise<boolean>
  • database - name of a database to be checked

Checks if database with the given name exist.


hasSchema(schema: string): Promise<boolean>
  • schema - name of a schema to be checked

Checks if schema with the given name exist. Used only for SqlServer and Postgres.


hasTable(table: Table|string): Promise<boolean>
  • table - Table object or name

Checks if table exist.


hasColumn(table: Table|string, columnName: string): Promise<boolean>
  • table - Table object or name
  • columnName - name of a column to be checked

Checks if column exist in the table.


createDatabase(database: string, ifNotExist?: boolean): Promise<void>
  • database - database name
  • ifNotExist - skips creation if true, otherwise throws error if database already exist

Creates a new database.


dropDatabase(database: string, ifExist?: boolean): Promise<void>
  • database - database name
  • ifExist - skips deletion if true, otherwise throws error if database was not found

Drops database.


createSchema(schemaPath: string, ifNotExist?: boolean): Promise<void>
  • schemaPath - schema name. For SqlServer can accept schema path (e.g. 'dbName.schemaName') as parameter. If schema path passed, it will create schema in specified database
  • ifNotExist - skips creation if true, otherwise throws error if schema already exist

Creates a new table schema.


dropSchema(schemaPath: string, ifExist?: boolean, isCascade?: boolean): Promise<void>
  • schemaPath - schema name. For SqlServer can accept schema path (e.g. 'dbName.schemaName') as parameter. If schema path passed, it will drop schema in specified database
  • ifExist - skips deletion if true, otherwise throws error if schema was not found
  • isCascade - If true, automatically drop objects (tables, functions, etc.) that are contained in the schema. Used only in Postgres.

Drops a table schema.


createTable(table: Table, ifNotExist?: boolean, createForeignKeys?: boolean, createIndices?: boolean): Promise<void>
  • table - Table object.
  • ifNotExist - skips creation if true, otherwise throws error if table already exist. Default false
  • createForeignKeys - indicates whether foreign keys will be created on table creation. Default true
  • createIndices - indicates whether indices will be created on table creation. Default true

Creates a new table.


dropTable(table: Table|string, ifExist?: boolean, dropForeignKeys?: boolean, dropIndices?: boolean): Promise<void>
  • table - Table object or table name to be dropped
  • ifExist - skips dropping if true, otherwise throws error if table does not exist
  • dropForeignKeys - indicates whether foreign keys will be dropped on table deletion. Default true
  • dropIndices - indicates whether indices will be dropped on table deletion. Default true

Drops a table.


renameTable(oldTableOrName: Table|string, newTableName: string): Promise<void>
  • oldTableOrName - old Table object or name to be renamed
  • newTableName - new table name

Renames a table.


addColumn(table: Table|string, column: TableColumn): Promise<void>
  • table - Table object or name
  • column - new column

Adds a new column.


addColumns(table: Table|string, columns: TableColumn[]): Promise<void>
  • table - Table object or name
  • columns - new columns

Adds a new column.


renameColumn(table: Table|string, oldColumnOrName: TableColumn|string, newColumnOrName: TableColumn|string): Promise<void>
  • table - Table object or name
  • oldColumnOrName - old column. Accepts TableColumn object or column name
  • newColumnOrName - new column. Accepts TableColumn object or column name

Renames a column.


changeColumn(table: Table|string, oldColumn: TableColumn|string, newColumn: TableColumn): Promise<void>
  • table - Table object or name
  • oldColumn - old column. Accepts TableColumn object or column name
  • newColumn - new column. Accepts TableColumn object

Changes a column in the table.


changeColumns(table: Table|string, changedColumns: { oldColumn: TableColumn, newColumn: TableColumn }[]): Promise<void>
  • table - Table object or name
  • changedColumns - array of changed columns.
    • oldColumn - old TableColumn object
    • newColumn - new TableColumn object

Changes a columns in the table.


dropColumn(table: Table|string, column: TableColumn|string): Promise<void>
  • table - Table object or name
  • column - TableColumn object or column name to be dropped

Drops a column in the table.


dropColumns(table: Table|string, columns: TableColumn[]|string[]): Promise<void>
  • table - Table object or name
  • columns - array of TableColumn objects or column names to be dropped

Drops a columns in the table.


createPrimaryKey(table: Table|string, columnNames: string[]): Promise<void>
  • table - Table object or name
  • columnNames - array of column names which will be primary

Creates a new primary key.


updatePrimaryKeys(table: Table|string, columns: TableColumn[]): Promise<void>
  • table - Table object or name
  • columns - array of TableColumn objects which will be updated

Updates composite primary keys.


dropPrimaryKey(table: Table|string): Promise<void>
  • table - Table object or name

Drops a primary key.


createUniqueConstraint(table: Table|string, uniqueConstraint: TableUnique): Promise<void>
  • table - Table object or name
  • uniqueConstraint - TableUnique object to be created

Creates new unique constraint.

Note: does not work for MySQL, because MySQL stores unique constraints as unique indices. Use createIndex() method instead.


createUniqueConstraints(table: Table|string, uniqueConstraints: TableUnique[]): Promise<void>
  • table - Table object or name
  • uniqueConstraints - array of TableUnique objects to be created

Creates new unique constraints.

Note: does not work for MySQL, because MySQL stores unique constraints as unique indices. Use createIndices() method instead.


dropUniqueConstraint(table: Table|string, uniqueOrName: TableUnique|string): Promise<void>
  • table - Table object or name
  • uniqueOrName - TableUnique object or unique constraint name to be dropped

Drops an unique constraint.

Note: does not work for MySQL, because MySQL stores unique constraints as unique indices. Use dropIndex() method instead.


dropUniqueConstraints(table: Table|string, uniqueConstraints: TableUnique[]): Promise<void>
  • table - Table object or name
  • uniqueConstraints - array of TableUnique objects to be dropped

Drops an unique constraints.

Note: does not work for MySQL, because MySQL stores unique constraints as unique indices. Use dropIndices() method instead.


createCheckConstraint(table: Table|string, checkConstraint: TableCheck): Promise<void>
  • table - Table object or name
  • checkConstraint - TableCheck object

Creates new check constraint.

Note: MySQL does not support check constraints.


createCheckConstraints(table: Table|string, checkConstraints: TableCheck[]): Promise<void>
  • table - Table object or name
  • checkConstraints - array of TableCheck objects

Creates new check constraint.

Note: MySQL does not support check constraints.


dropCheckConstraint(table: Table|string, checkOrName: TableCheck|string): Promise<void>
  • table - Table object or name
  • checkOrName - TableCheck object or check constraint name

Drops check constraint.

Note: MySQL does not support check constraints.


dropCheckConstraints(table: Table|string, checkConstraints: TableCheck[]): Promise<void>
  • table - Table object or name
  • checkConstraints - array of TableCheck objects

Drops check constraints.

Note: MySQL does not support check constraints.


createForeignKey(table: Table|string, foreignKey: TableForeignKey): Promise<void>
  • table - Table object or name
  • foreignKey - TableForeignKey object

Creates a new foreign key.


createForeignKeys(table: Table|string, foreignKeys: TableForeignKey[]): Promise<void>
  • table - Table object or name
  • foreignKeys - array of TableForeignKey objects

Creates a new foreign keys.


dropForeignKey(table: Table|string, foreignKeyOrName: TableForeignKey|string): Promise<void>
  • table - Table object or name
  • foreignKeyOrName - TableForeignKey object or foreign key name

Drops a foreign key.


dropForeignKeys(table: Table|string, foreignKeys: TableForeignKey[]): Promise<void>
  • table - Table object or name
  • foreignKeys - array of TableForeignKey objects

Drops a foreign keys.


createIndex(table: Table|string, index: TableIndex): Promise<void>
  • table - Table object or name
  • index - TableIndex object

Creates a new index.


createIndices(table: Table|string, indices: TableIndex[]): Promise<void>
  • table - Table object or name
  • indices - array of TableIndex objects

Creates a new indices.


dropIndex(table: Table|string, index: TableIndex|string): Promise<void>
  • table - Table object or name
  • index - TableIndex object or index name

Drops an index.


dropIndices(table: Table|string, indices: TableIndex[]): Promise<void>
  • table - Table object or name
  • indices - array of TableIndex objects

Drops an indices.


clearTable(tableName: string): Promise<void>
  • tableName - table name

Clears all table contents.

Note: this operation uses SQL's TRUNCATE query which cannot be reverted in transactions.


enableSqlMemory(): void

Enables special query runner mode in which sql queries won't be executed, instead they will be memorized into a special variable inside query runner. You can get memorized sql using getMemorySql() method.


disableSqlMemory(): void

Disables special query runner mode in which sql queries won't be executed. Previously memorized sql will be flushed.


clearSqlMemory(): void

Flushes all memorized sql statements.


getMemorySql(): SqlInMemory
  • returns SqlInMemory object with array of upQueries and downQueries sql statements

Gets sql stored in the memory. Parameters in the sql are already replaced.


executeMemoryUpSql(): Promise<void>

Executes memorized up sql queries.


executeMemoryDownSql(): Promise<void>

Executes memorized down sql queries.