Schema-less in MongoDB

Although MongoDB does not enforce a complete schema, it is not entirely schema-free. You can define validation rules to enforce a minimal structure on the documents within a collection.

This is useful when you want flexibility, but also need to ensure that certain fields always exist or follow specific data types.

JSON Schema Validation

MongoDB supports schema validation using the $jsonSchema keyword. This allows you to describe the expected structure of documents in a way that is very similar to the JSON Schema Standard.

When you create a collection, you can attach a validator that checks each document being inserted or updated. The following example shows how to create a products collection in MongoDB with a JSON Schema validator.

db.createCollection("products", {
  validator: {
    $jsonSchema: {
      bsonType: "object",
      required: ["name", "price"],
      properties: {
        name: {
          bsonType: "string",
          description: "Product name must be a string and is required."
        },
        price: {
          bsonType: "number",
          minimum: 0,
          description: "Price must be a positive number."
        },
        specs: {
          bsonType: "object",
          properties: {
            wireless: {
              bsonType: "bool",
              description: "Indicates whether the product is wireless."
            }
          }
        }
      }
    }
  },
  validationAction: "error", // reject invalid documents
  validationLevel: "strict"  // Validate all inserts and updates
})

With the above JSON Schema definition a document must include the name and price properties as they are listed in the required array. The specs property is optional.

The name property must be a string;
The price property must be a number with a minimum value of zero;
If present, the specs property must be an object;
If the specs object contains the wireless property it must be a boolean.

After creating the collection above:

The following document will be accepted:

db.products.insertOne({
  name: "Headphones",
  price: 199,
  specs: { wireless: true }
})

The following document will be rejected (missing price):

db.products.insertOne({
  name: "Keyboard",
  specs: { wireless: false }
})

Validation Options

MongoDB allows you to define validation behaviour for documents using two options:

validationAction

“error” → Invalid documents are rejected and not written to the collection (default).
“warn” → invalid documents are still written, but a warning is logged.

validationLevel

“strict” → All inserts and updates are validated against the schema (default).
“moderate” → Inserts are always validated. Updates are only validated if the existing document already meets the validation rules.

Adding a Validator to an Existing Collection

If the collection already exists, you can still apply a JSON Schema validator afterwards. This is done with the collMod (collection modify) command.

db.runCommand({
  collMod: "products",
  validator: {
    $jsonSchema: {
      bsonType: "object",
      required: ["name", "price"],
      properties: {
        name: {
          bsonType: "string",
          description: "Product name must be a string and is required."
        },
        price: {
          bsonType: "number",
          minimum: 0,
          description: "Price must be a positive number."
        },
        specs: {
          bsonType: "object",
          properties: {
            wireless: {
              bsonType: "bool",
              description: "Indicates whether the product is wireless."
            }
          }
        }
      }
    }
  },
  validationAction: "error",
  validationLevel: "strict"  // Validate all inserts and updates
})

This command updates the existing products collection and applies the same validation rules as if they had been defined at creation time.

Validation of existing Documents

When you add a validator to a collection that already contains documents, MongoDB does not check the existing documents immediately. Validation is only enforced when documents are inserted or updated.