但也可以使用其它备选 `servers`，例如，需要同一个 API 文档与 staging 和生产环境交互。

如果传递自定义 `servers` 列表，并有 `root_path`（ 因为 API 使用了代理），**FastAPI** 会在列表开头使用这个 `root_path` 插入**服务器**。

例如：

{* ../../docs_src/behind_a_proxy/tutorial003.py hl[4:7] *}

这段代码生产如下 OpenAPI 概图：

```JSON hl_lines="5-7"
{
    "openapi": "3.0.2",
    // More stuff here
    "servers": [
        {
            "url": "/api/v1"

        return 0;
    }

    @Override
    int readParametersWireFormat(final byte[] buffer, final int bufferIndex, final int len) {
        // observed two zero bytes here with at least win98
        return 2;
    }

    @Override
    int readDataWireFormat(final byte[] buffer, final int bufferIndex, final int len) {
        return switch (informationLevel) {

 * Client and server exchange these certificates during the handshake phase of a TLS connection.
 *
 * ### Server Authentication
 *
 * This is the most common form of TLS authentication: clients verify that servers are trusted and
 * that they own the hostnames that they represent. Server authentication is required.
 *
 * To perform server authentication:
 *

- MinIO exports Prometheus compatible data by default which is bucket centric as an authorized endpoint at `/minio/v2/metrics/bucket`.

This document explains how to setup Prometheus and configure it to scrape data from MinIO servers.

## Prerequisites

To get started with MinIO, refer [MinIO QuickStart Document](https://docs.min.io/community/minio-object-store/operations/deployments/baremetal-deploy-minio-on-redhat-linux.html).

## Prometheus Probe

MinIO allows reading metrics for the entire cluster from any single node. This allows for metrics collection for a MinIO instance across all servers. Thus, metrics collection for instances behind a load balancer can be done without any knowledge of the individual node addresses. The cluster wide metrics can be read at
`<Address for MinIO Service>/minio/v2/metrics/cluster`.

import org.opentest4j.TestAbortedException

/**
 * This test binds two different web servers (IPv4 and IPv6) to the same port, but on different
 * local IP addresses. Requests made to `127.0.0.1` will reach the IPv4 server, and requests made to
 * `::1` will reach the IPv6 server.
 *
 * By orchestrating two different servers with the same port but different IP addresses, we can

	# Wait for all drives to be online and formatted
	while [ $(/tmp/mc admin info --json myminio | jq '.info.servers[].drives[].state | select(. != "ok")' | wc -l) -gt 0 ]; do sleep 1; done
	# Wait for all drives to be healed
	while [ $(/tmp/mc admin info --json myminio | jq '.info.servers[].drives[].healing | select(. != null) | select(. == true)' | wc -l) -gt 0 ]; do sleep 1; done

	# Wait for Status: in MinIO output

          .url(server.url("/"))
          .build(),
      )
    call1.execute().use { response -> response.body.string() }

    listener.shouldClose = true
    // Large headers are a likely reason the servers would cut off the connection before it completes sending
    // request headers.
    // 431 "Request Header Fields Too Large"
    val largeHeaders =
      Headers
        .Builder()
        .apply {

   *       {@code consistentHash} will handle it. If, however, you are dividing traffic among
   *       servers {@code alpha}, {@code bravo}, and {@code charlie} and you occasionally need to
   *       take each of the servers offline, {@code consistentHash} will be a poor fit: It provides
   *       no way for you to specify which of the three buckets is disappearing. Thus, if your

     * This filter can be considerably more efficient than other file filters
     * as the specified wildcard and attributes are passed to the server for
     * filtering there (although attributes are largely ignored by servers
     * they are filtered locally by the default accept method).
     *
     * @param wildcard the wildcard pattern for matching file names (e.g., "*.txt")