> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vmeg.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Request ID

> Correlate Open API calls with req_id for logging and support

## Why use it

When a call fails or you need help from VMEG support, a **request ID** ties together server logs, idempotency records, and task creation for that HTTP request.

You can pass your own ID on **any** Open API request (`/openapi/v1/**`). If you omit it, the server generates one (32-character hex, no hyphens).

<Info>
  `req_id` marks **this HTTP request** for logging and support. [`X-Idempotency-Key`](/guides/idempotency) prevents duplicate writes, `taskId` tracks an async job, and [`pipelineKey`](/guides/webhook-request#deduplicate-by-pipelinekey) deduplicates webhook deliveries — use the field that matches your goal; they are often used together.
</Info>

## What to send

Add a query parameter:

| Item     | Value        |
| -------- | ------------ |
| Name     | `req_id`     |
| Location | Query string |
| Required | No           |

```http theme={null}
POST /openapi/v1/task/tts/create?req_id=YOUR_REQUEST_ID
```

Rules:

* Use a value you can store in your own logs (for example a UUID or `order-12345-attempt-1`)
* Works on `GET` and `POST` under `/openapi/v1/**`
* The ID is **not** returned in the standard success JSON body; keep the value you sent (or log the generated ID from your client if you rely on server generation only indirectly via support lookup)

## Example

```bash theme={null}
export REQ_ID="support-$(date +%s)-$RANDOM"

curl -X POST "https://api.vmeg.ai/openapi/v1/task/tts/create?req_id=${REQ_ID}" \
  -H "Authorization: Bearer $API_KEY" \
  -H "X-Idempotency-Key: $IDEM_KEY" \
  -H "Content-Type: application/json" \
  -d @payload.json
```

If the URL already has query parameters, append with `&req_id=...`.

## When contacting support

Provide:

* The `req_id` you used (or the approximate time and which **API Key** you used if you did not set one)
* `taskId` for async jobs, if applicable
* Product and endpoint path

<Tip>
  Set `req_id` on the **first** attempt of a flow and reuse it only when retrying the **same** HTTP call for debugging. Start a **new** `req_id` for a logically new submission (and use a new idempotency key for new writes).
</Tip>