Idempotency

Advanced: Idempotency

Idempotency ensures that making the same API request multiple times produces the same result as making it once. This is crucial for building robust applications that can safely retry requests after network errors or timeouts without causing unintended side effects (like creating duplicate resources or charging a user multiple times).

How it Works

REALM AI supports idempotency for specific mutating operations (like creating models, listing items on the marketplace, initiating purchases) via the Idempotency-Key HTTP header.

Generate a Key: When making a potentially mutating request for the first time, your client application generates a unique key (e.g., a UUID v4). This key should be unique for each distinct operation you want to perform.
Send the Key: Include this unique key in the Idempotency-Key header of your API request.
Server Processing:
- The first time the server receives a request with a specific Idempotency-Key, it processes the request as usual and stores the result (including the status code and response body) associated with that key.
- If the server receives another request with the same Idempotency-Key within a certain time window (e.g., 24 hours), it will not re-process the request. Instead, it will return the cached result from the original request.

Using Idempotency Keys

Example Scenario: Listing an item

You want to list model-123 for sale. Your application generates a unique key: uuid-abc-123.

Request 1:

POST /v1/marketplace/listings HTTP/1.1
Host: api.realm.ai
X-API-Key: YOUR_API_KEY
Idempotency-Key: uuid-abc-123
Content-Type: application/json
...

{ "assetId": "model-123", ... }

The server processes this, creates the listing, and returns a 201 Created response.

Request 2 (e.g., after a network timeout and retry):
```
POST /v1/marketplace/listings HTTP/1.1
Host: api.realm.ai
X-API-Key: YOUR_API_KEY
Idempotency-Key: uuid-abc-123 // SAME KEY
Content-Type: application/json
...

{ "assetId": "model-123", ... }
```
The server recognizes uuid-abc-123, sees it was already processed successfully, and immediately returns the same 201 Created response from the first request without creating a second listing.

Best Practices

Use for Mutating Requests: Apply idempotency keys primarily to POST, PUT, PATCH, and DELETE requests where unintended duplication could be harmful.
Generate Unique Keys: Ensure your key generation mechanism produces highly unique keys (UUIDs are recommended).
Key Scope: Use the same key only when retrying the exact same logical operation. Use a new key for a distinct operation (e.g., listing a different model).
Storage: Your client might need to store the generated idempotency key alongside the operation details in case a retry is needed later.
Expiration: Idempotency keys typically expire on the server after a period (e.g., 24 hours). Retrying with the same key after expiration will likely be treated as a new request.

PreviousUsers NextRate Limiting

Last updated 1 month ago

Advanced: Idempotency

How it Works

REALM AI supports idempotency for specific mutating operations (like creating models, listing items on the marketplace, initiating purchases) via the Idempotency-Key HTTP header.

Generate a Key: When making a potentially mutating request for the first time, your client application generates a unique key (e.g., a UUID v4). This key should be unique for each distinct operation you want to perform.

Send the Key: Include this unique key in the Idempotency-Key header of your API request.

Server Processing:

The first time the server receives a request with a specific Idempotency-Key, it processes the request as usual and stores the result (including the status code and response body) associated with that key.
If the server receives another request with the same Idempotency-Key within a certain time window (e.g., 24 hours), it will not re-process the request. Instead, it will return the cached result from the original request.

Using Idempotency Keys

Example Scenario: Listing an item

You want to list model-123 for sale. Your application generates a unique key: uuid-abc-123.

Request 1:

POST /v1/marketplace/listings HTTP/1.1
Host: api.realm.ai
X-API-Key: YOUR_API_KEY
Idempotency-Key: uuid-abc-123
Content-Type: application/json
...

{ "assetId": "model-123", ... }

The server processes this, creates the listing, and returns a 201 Created response.

Request 2 (e.g., after a network timeout and retry):

POST /v1/marketplace/listings HTTP/1.1
Host: api.realm.ai
X-API-Key: YOUR_API_KEY
Idempotency-Key: uuid-abc-123 // SAME KEY
Content-Type: application/json
...

{ "assetId": "model-123", ... }

The server recognizes uuid-abc-123, sees it was already processed successfully, and immediately returns the same 201 Created response from the first request without creating a second listing.

Best Practices

Use for Mutating Requests: Apply idempotency keys primarily to POST, PUT, PATCH, and DELETE requests where unintended duplication could be harmful.

Generate Unique Keys: Ensure your key generation mechanism produces highly unique keys (UUIDs are recommended).

Key Scope: Use the same key only when retrying the exact same logical operation. Use a new key for a distinct operation (e.g., listing a different model).

Storage: Your client might need to store the generated idempotency key alongside the operation details in case a retry is needed later.

Expiration: Idempotency keys typically expire on the server after a period (e.g., 24 hours). Retrying with the same key after expiration will likely be treated as a new request.