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

# Restart PIN Training Agent Process

> Restart a PIN training agent process on its GPU node without redeploying its binary or losing its registration.

Restart the agent process without redeploying the binary. Useful after configuration changes or when the process is stuck. The agent re-registers with the control plane on start.

## Endpoint

```http theme={null}
POST /v1/agents/{id}/restart
```

## Authentication

`Authorization: Bearer <api_key or JWT>` with fleet management rights on the target agent.

## Path parameters

<ParamField path="id" type="string" required>
  Agent identifier.
</ParamField>

## Response

<ResponseField name="code" type="integer">
  `0` on accepted restart request.
</ResponseField>

<ResponseField name="data" type="object">
  Command envelope; typically `{ "ok": true, "command_id": "..." }` for the queued restart command.
</ResponseField>

## Example

```bash theme={null}
curl -X POST https://ariapin.example.com/v1/agents/ag_01H7Z.../restart \
  -H "Authorization: Bearer $ARIAPIN_API_KEY"
```

```json theme={null}
{
  "code": 0,
  "data": { "ok": true, "command_id": "cmd_9f8..." },
  "message": ""
}
```

<Note>
  The restart is asynchronous. Poll the agent status endpoint to confirm it has come back online.
</Note>


## Related topics

- [Reload a model into a serving runtime](/api-reference/pin/models/reload.md)
- [Node Agent Control Channel and Interventions](/api-reference/pin/node-agent/control.md)
- [Report Training Metrics and Status - Node Agent](/api-reference/pin/node-agent/metrics.md)
- [Merge LoRA adapter into a base model](/api-reference/pin/models/merge.md)
- [Export a trained model to GGUF, MLX, MNN, or QNN](/api-reference/pin/models/export.md)