Prisma FAQ & Answers

TypeScript 4.7 or higher.

"postinstall": "prisma generate" in the scripts section. This ensures Prisma Client is regenerated at build time, solving caching issues on deployment platforms.

PrismaClient automatically connects lazily when you run your first query. It creates a connection pool at that time and disconnects when the Node.js process ends.

Prisma looks for schemas in two default locations: ./prisma/schema.prisma and ./schema.prisma.

Organize your files by domain: group related models into the same file. For example, keep all user-related models in user.prisma while post-related models go in post.prisma.

When using a multi-file Prisma schema, you must always explicitly specify the location of the directory that contains your schema files (including the main schema.prisma file with your generator block) using --schema flag, package.json configuration, or prisma.config.ts.

Prisma Studio currently supports PostgreSQL, MySQL, and SQLite. Support for CockroachDB and MongoDB is not currently available.

Neon Serverless, PlanetScale Serverless, and @libsql/client (for Turso) work with both Cloudflare Workers and Vercel Edge Functions. The node-postgres (pg) driver with Cloudflare's connect() only works with Cloudflare Workers, not Vercel Edge Functions.

As of Prisma ORM v7, the url, directUrl, and shadowDatabaseUrl fields in the Prisma schema datasource block are deprecated. These fields should be configured in Prisma Config instead.

TypedSQL .sql files should be placed in the prisma/sql directory.

MongoDB raw query support ($runCommandRaw, findRaw, and aggregateRaw) was added in Prisma version 3.9.0.

By default, Prisma ORM's String type gets mapped to MySQL's varchar type.

When using scalar list filters with relational database connectors, array fields with a NULL value are not considered by conditions like NOT (array does not contain X) and isEmpty (array is empty).

Yes, you can nest select inside an include to have fine-grained control over which fields are returned from a relation. Example: include: { posts: { select: { title: true } } }

The omit feature was released into General Availability with Prisma ORM 6.2.0. It was previously available via the omitApi Preview feature in versions 5.13.0 through 6.1.0.

If you pass an empty update clause, the @updatedAt value will remain unchanged.

If you do not specify referential actions, Prisma ORM uses Restrict as the default for onDelete and Cascade as the default for onUpdate.

Prisma ORM v7 uses a TypeScript-based query compiler with WebAssembly by default, no longer depending on Rust native binaries. This enables use in environments that support JavaScript or WASM, such as Cloudflare Workers, Bun, and Deno.

For Prisma v6.19 and earlier with Rust-based engines, you can use library (default, uses Node-API) or binary (uses executable binary). In v6.16+, you can use client (Generally Available) to run without Rust binaries using driver adapters.

Each model can have a maximum of one @id or @@id for a single- or multi-field primary key constraint per model.

The five referential actions are: Cascade (deletes/updates related records), Restrict (prevents operation if records are connected, not available on SQL Server), SetNull (sets fields to null), SetDefault (sets fields to default values, not supported on MySQL/MariaDB), and NoAction (database-specific behavior, may defer integrity checks).

UserDefaultArgs is used with Prisma.validator to create type-safe objects for select or include operations, like: Prisma.validator<Prisma.UserDefaultArgs>()({ include: { posts: true } })

By default, Prisma Client is generated into the node_modules/.prisma/client folder. The actual types are stored in the .prisma/client folder, and @prisma/client/index.d.ts exports the contents of this folder.

The rejectOnNotFound parameter was deprecated in Prisma version 4.0.0.

The default value for maxWait is 2000 milliseconds (2 seconds).

Add the mode property with value 'insensitive' to the filter, for example: { email: { endsWith: 'prisma.io', mode: 'insensitive' } }

No, Prisma Client's distinct option does not use SQL SELECT DISTINCT. It was designed this way to support select and include as part of distinct queries.

P2003

Failed to validate the query

The default selection set includes all scalar fields defined in the Prisma schema (including enums and arrays/scalar lists), but none of the relation fields. Relations are not included by default because they are not scalar fields.

Yes. You can map String fields to PostgreSQL's UUID type using @db.Uuid, commonly with database-generated defaults like @default(dbgenerated('gen_random_uuid()'))

In Prisma 5+, the path argument for PostgreSQL only accepts an array format, for example: path: ['someSetting'] or path: ['pet2', 'petName']

Sets relation scalar fields to their default values on update or delete of relation.

PostgreSQL supports 'and' (&) and 'or' (|) operators for combining search terms.

Prisma Migrate requires that the database user defined in your datasource has permission to create databases in order to automatically create and delete the shadow database.

In the event of a Rust panic from the Prisma engine, the CLI exits the process with a non-successful exit code of 1. Previously, Prisma ended the process with a successful exit code of 0.

If you introspect a database, the names for indexes and constraints will be added to your schema using the map argument unless they follow Prisma ORM's naming convention - if they do, the names are not rendered to keep the schema more readable.

No, there are known issues where Prisma does not use LIMIT in the generated SQL for certain cursor-based pagination scenarios, which can cause performance issues and memory problems.

ASC (ascending) order is the default when no sort direction is specified.

Static default values for DateTime are based on the ISO 8601 standard. Example: data DateTime @default("2020-03-19T14:21:00+02:00")

Use Prisma.defineExtension() to make extensions shareable. This provides strict type checks and auto completion for both extension authors and users.

Middleware executes for every query, which means that overuse has the potential to negatively impact performance. To avoid adding performance overheads, check the params.model and params.action properties early in your middleware to avoid running logic unnecessarily.

No, case-insensitive sorting is not currently supported in orderBy. The 'mode' parameter (used for case-insensitive filtering) is not available for orderBy. Sorting case sensitivity is determined by database collation.

Yes, Pulse will store the events that happen in your database in the same shape that they're delivered. This allows you to resume streams with zero data loss even when your server goes down.

Prisma recommends enabling the TypeScript compiler option exactOptionalPropertyTypes alongside strictUndefinedChecks, which will help catch cases of undefined values at compile time.

No. The @id attribute can only be defined on scalar fields (String, Int, enum), not on relation fields.

If you don't provide a value to the nanoid() function, the default value is 21. It accepts an integer value between 2 and 255 that specifies the length of the generated ID value.

No, once the wal_level setting is changed to logical, it cannot be reverted.

Use the @@id attribute with an array of field names at the model level. For example: @@id([firstName, lastName]) creates a composite primary key from both fields.

Filtering on object key values within an array is only supported by the MySQL database connector. PostgreSQL does not support this feature.

Use the PostgreSQL cast operator ::text. For example: await prisma.$queryRawSELECT location::text FROM Country;``

When run in Bun, the generated config does not include the import 'dotenv/config' line because Bun automatically loads .env files, whereas Node.js configurations include this import. The feature was introduced in Prisma ORM 7.2.0.

No. Prisma does not offer a manual transaction rollback API. To rollback an interactive transaction, you must throw an error, which triggers automatic rollback.

No. When you enable Accelerate and use the Accelerate connection string, your queries route through a connection pooler by default without the performance overhead of wrapping all queries in transactions.

When NODE_ENV is set to 'production', Prisma automatically uses the 'minimal' error format, which shows only the raw error message. This allows for easier digestion of logs in production environments.

prisma is the default and only available relation mode for MongoDB.

PostgreSQL supports ALTER TYPE example_enum RENAME VALUE 'FOO' TO 'BAR' (added in Postgres 10), but Prisma doesn't currently use this method automatically. Instead, it generates an unsafe migration that creates a new enum type and attempts to convert existing data.

When there are migration history conflicts caused by modified or missing migrations, or when schema drift is detected (database schema has diverged from migration history). A migration that was already applied but later modified will trigger this prompt.

The Prisma ORM quickstart with Prisma Postgres takes 5 minutes to complete.

Each migration in the _prisma_migrations table has a logs column that stores error messages when a migration fails. This can help with debugging failed migrations.

The 'type' keyword is used in the Prisma schema to define composite types.

TypedSQL requires Prisma ORM version 5.19.0 or later.

After adding @@fulltext to your schema, you must run prisma generate and either prisma db push or prisma migrate dev to update your database schema.

Yes, if you use skip and/or take with groupBy(), you must also include orderBy in the query.

In MySQL versions 8 and later, and MariaDB versions 10.5 and later, SetDefault effectively acts as an alias for NoAction.

The shadow database is a second, temporary database that is created and deleted automatically each time you run prisma migrate dev and is primarily used to detect problems such as schema drift or potential data loss of the generated migration.

If sanitization results in duplicate identifiers, no immediate error handling is in place. You get the error later when running prisma generate and must manually fix it by renaming one of the models.

No. In Prisma v6, full-text search was promoted to General Availability for MySQL. You can remove fullTextSearch from the previewFeatures in your Prisma schema if you use MySQL.

The Prisma Accelerate connection pooler is available in 16 regions globally.

For relational databases, Prisma uses the foreignKeys relation mode by default, which enforces relations between records at the database level with foreign keys.

Using engine type "client" requires either "adapter" or "accelerateUrl" to be provided to the PrismaClient constructor.

It is only possible to switch between relation modes when you use a relational database connector: PostgreSQL, MySQL, SQLite, SQL Server, or CockroachDB.

prisma (emulation mode)

The entire transaction is rolled back. If the application encounters an error along the way, the async function will throw an exception and automatically rollback the transaction.

No, the onDelete and onUpdate arguments must only be specified on the side of the relation that contains the foreign key (the side with the @relation attribute that has fields and references).

Install the Prisma ORM versions 2.x and later CLI and run npx prisma init. Then connect to your database and introspect it with npx prisma db pull.

In Prisma 5, when filtering on JSON fields in a PostgreSQL model, the path argument now only accepts an array. The syntax must be path: ['someSetting'] instead of the Prisma 4 syntax path: 'someSetting'.

In Prisma ORM v6, the default max_idle_connection_lifetime is 300 seconds (5 minutes). This is how long an idle connection can remain in the pool before being closed.

No. You can query models in multiple database schemas without any change to your Prisma Client query syntax.

Native database type attributes were introduced in Prisma version 2.17.0.

Next.js's hot-reloading feature in development mode can create multiple instances of Prisma Client, which consumes resources and might cause unexpected behavior. The singleton pattern stores PrismaClient on globalThis in development, which persists across hot reloads, ensuring only one instance exists.

Prisma recommends using the 'prisma db push' command instead of 'prisma migrate' when making schema changes with PlanetScale.

The provider value must be "mongodb". Example: datasource db { provider = "mongodb" url = env("DATABASE_URL") }

The Prisma Postgres free tier includes 100,000 operations per month.

CREATE PUBLICATION $PUBLICATION_NAME FOR ALL TABLES;

You should edit migration SQL when: (1) renaming fields or tables to avoid data loss, (2) changing the direction of a 1-1 relation without data loss, (3) introducing significant refactors, or (4) adding features that cannot be represented in Prisma Schema Language such as partial indexes, stored procedures, views, or triggers.

Prisma.AnyNull represents both JSON null values and database NULL values, and can only be used when filtering (not for creating or updating data). It matches either type of null value in queries.

@opentelemetry/api became a peer dependency instead of a regular dependency. Users must install it separately with npm install @opentelemetry/api.

is requires one or more fields to be present, while equals requires all fields to be present. Use is for partial matching and equals for exact matching.

No, using only false values (e.g., select: { password: false }) will cause an error requiring at least one true field. Prisma recommends using the omit option instead to exclude specific fields.

SetNull. This is the default when you don't explicitly specify onDelete in the @relation attribute.

import { PrismaMssql } from '@prisma/adapter-mssql'

const sqlConfig = {
  user: process.env.DB_USER,
  password: process.env.DB_PASSWORD,
  database: process.env.DB_NAME,
  server: process.env.HOST,
  pool: {
    max: 10,
    min: 0,
    idleTimeoutMillis: 30000
  },
  options: {
    encrypt: true,
    trustServerCertificate: false
  }
}

const adapter = new PrismaMssql(sqlConfig)
const prisma = new PrismaClient({ adapter })

No. The --skip-generate and --skip-seed flags were removed in Prisma 7 because migrate dev no longer runs prisma generate or seeds automatically.

The connection pool is created when Prisma Client opens the first connection to the database. This happens when the query engine is started, which occurs when the first Prisma Client query is invoked or when the $connect() method is called.

Yes. View fields support native database type attributes such as @db.VarChar(200) or @db.Text, which determine the specific native type in the database. These work the same way as they do for model fields.

Yes. Operations are executed sequentially in the order they appear in the array. Prisma provides a dedicated transaction API which ensures these operations are executed in order and are guaranteed to either succeed or fail as a whole.

The command is prisma db pull. It connects to your database and adds Prisma models to your Prisma schema that reflect the current database schema.

No, 'datasource db' is just a convention. You can give your data source any name, for example, 'datasource mysql' or 'datasource data'.

Offset pagination uses skip and take parameters. The skip parameter specifies how many records to skip, and take specifies how many records to return.

February 16, 2024. Since then, you can alternatively use foreign key constraints at the database level in PlanetScale, which eliminates the need for setting relationMode = 'prisma'.