---

## Instale o FastAPI

O primeiro passo é instalar o FastAPI.

Para o tutorial, você deve querer instalá-lo com todas as dependências e recursos opicionais.

<div class="termy">

```console
$ pip install "fastapi[standard]"

---> 100%
```

</div>

/// note | Nota

	apiRejectedInvalidTotalMD = NewCounterMD(apiRejectedInvalidTotal,
		"Total number of invalid requests", "type")

	apiRequestsWaitingTotalMD = NewGaugeMD(apiRequestsWaitingTotal,
		"Total number of requests in the waiting queue", "type")
	apiRequestsIncomingTotalMD = NewGaugeMD(apiRequestsIncomingTotal,
		"Total number of incoming requests", "type")

	apiRequestsInFlightTotalMD = NewGaugeMD(apiRequestsInFlightTotal,

Por ejemplo, podrías decidir leer y validar el request con tu propio código, sin usar las funcionalidades automáticas de FastAPI con Pydantic, pero aún podrías querer definir el request en el esquema de OpenAPI.

Podrías hacer eso con `openapi_extra`:

{* ../../docs_src/path_operation_advanced_configuration/tutorial006.py hl[19:36, 39:40] *}

     *
     * <ol>
     *   <li>Update the {@link #servicesByState()}
     *   <li>Update the {@link #startupTimers}
     *   <li>Based on the new state queue listeners to run
     *   <li>Run the listeners (outside of the lock)
     * </ol>
     */
    void transitionService(Service service, State from, State to) {
      checkNotNull(service);

     * Get the value for the key 'logging.search.max.queue.size'. <br>
     * The value is, e.g. 10000 <br>
     * comment: Maximum queue size for search logging.
     * @return The value of found property. (NotNull: if not found, exception but basically no way)
     */
    String getLoggingSearchMaxQueueSize();

    /**
     * Get the value for the key 'logging.search.max.queue.size' as {@link Integer}. <br>

### Contenedores con Múltiples Procesos y Casos Especiales

Por supuesto, hay **casos especiales** donde podrías querer tener **un contenedor** con varios **worker processes de Uvicorn** dentro.

En esos casos, puedes usar la opción de línea de comandos `--workers` para establecer el número de workers que deseas ejecutar:

      server.enqueue(
        MockResponse
          .Builder()
          .body("enqueued in the background")
          .build(),
      )
    }.start()
    val connection = server.url("/").toUrl().openConnection()
    val reader = BufferedReader(InputStreamReader(connection!!.getInputStream(), UTF_8))
    assertThat(reader.readLine()).isEqualTo("enqueued in the background")
  }

  @Test
  fun nonHexadecimalChunkSize() {

    }

    assertThat(server.requestCount).isEqualTo(0)

    // Confirm that the connection pool was not corrupted by making another call. This doesn't use
    // makeSimpleCall() because it uses the MockResponse enqueued above.
    val callB = client.newCall(Request(server.url("/")))
    callB.execute().use { response ->
      assertThat(response.body.string()).isEqualTo("abc")
    }
  }

  private fun makeSimpleCall() {

    }

    /**
     * Returns true if the request contains conditions that save the server from sending a response
     * that the client has locally. When a request is enqueued with its own conditions, the built-in
     * response cache won't be used.
     */
    private fun hasConditions(request: Request): Boolean =

///

### `startup` y `shutdown` juntos

Hay una gran posibilidad de que la lógica para tu *startup* y *shutdown* esté conectada, podrías querer iniciar algo y luego finalizarlo, adquirir un recurso y luego liberarlo, etc.