IdentityServer4环境部署失败分析贴（一）

前言：

在部署Idv4站点和其客户端在外网时，发现了许多问题，折腾了许久，翻看了许多代码，写个MD记录一下。

1.受保护站点提示错误： Unable to obtain configuration from: '[PII is hidden]'.

fail: Microsoft.AspNetCore.Server.Kestrel[13]
      Connection id "0HLL3MB34N0G5", Request id "0HLL3MB34N0G5:00000009": An unhandled exception was thrown by the application.
System.InvalidOperationException: IDX20803: Unable to obtain configuration from: '[PII is hidden]'.
   在 Microsoft.IdentityModel.Protocols.ConfigurationManager`1.<GetConfigurationAsync>d__24.MoveNext()
--- 引发异常的上一位置中堆栈跟踪的末尾 ---
   在 System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw()
   在 System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   在 Microsoft.AspNetCore.Authentication.OpenIdConnect.OpenIdConnectHandler.<HandleChallengeAsync>d__18.MoveNext()
--- 引发异常的上一位置中堆栈跟踪的末尾 ---
   在 System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw()
   在 System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   在 Microsoft.AspNetCore.Authentication.AuthenticationHandler`1.<ChallengeAsync>d__54.MoveNext()
--- 引发异常的上一位置中堆栈跟踪的末尾 ---

1.1 受保护站点的代码执行流程

graph TB
User--1.进入-->A(Home/Index)
A--2.Authorize-->AA{是否授权}
AA--YES-->完事
AA--NO-->B(返回ChallengeResult)
B--3.ExecuteResult-->C(进入AuthenticationMiddleware中间件)
C--4.根据authenticationScheme找到相应Handler-->D(OpenIDConenctHandler)
D--5.调用HandleChallengeAsync-->F(调用OpenIdConnectConfigurationRetriever.IConfigurationRetriever)
F--6.通过HTTP调用Idv4站点的/.well-known/openid-configuration-->F1(Idv4Config)
F1--7.通过上步骤取到的config.jwks_uri获取加密key信息-->F2(获得公钥私钥等信息)
F2--8.设置JsonWebKeySet-->F3(完成openIdConnectConfiguration.SigningKeysKeys的添加)
F3-->F4(完成)

1.1 受保护站点的部分源码

Idv4部分源码

//注入Idev相关，配置authenticationScheme为oidc
public static IIdentityServerBuilder AddIdentityServer(this IServiceCollection services, Action<IdentityServerOptions> setupAction)
{
    services.Configure(setupAction);
    return services.AddIdentityServer();
}

PS: 这里要额外提一下，在Abp的模板项目中，如下配置使用Idv4可能无法正常跳转IDV4授权站完成单点登录

services.AddOpenIdConnect("oidc", options =>
{
   ...
}

需要额外添加下面这段，原因是Abp引入Microsoft.Identity,会重复设置schme导致错误，需如下代码添加设置。

services.AddAuthentication(options =>
{
        options.DefaultAuthenticateScheme = "oidc";
        options.DefaultChallengeScheme = "oidc";
        options.DefaultSignOutScheme = "oidc";
})

步骤5部分源码

 protected override async Task HandleChallengeAsync(AuthenticationProperties properties)
        {
            Logger.EnteringOpenIdAuthenticationHandlerHandleUnauthorizedAsync(GetType().FullName);

            // order for local RedirectUri
            // 1. challenge.Properties.RedirectUri
            // 2. CurrentUri if RedirectUri is not set)
            if (string.IsNullOrEmpty(properties.RedirectUri))
            {
                properties.RedirectUri = CurrentUri;
            }
            Logger.PostAuthenticationLocalRedirect(properties.RedirectUri);

            if (_configuration == null && Options.ConfigurationManager != null)
            {
                //问题在这里
                _configuration = await Options.ConfigurationManager.GetConfigurationAsync(Context.RequestAborted);
            }
            ...

步骤6部分源码

 public static async Task<OpenIdConnectConfiguration> GetAsync(string address, IDocumentRetriever retriever, CancellationToken cancel)
        {
            if (string.IsNullOrWhiteSpace(address))
                throw LogHelper.LogArgumentNullException(nameof(address));

            if (retriever == null)
            {
                throw LogHelper.LogArgumentNullException(nameof(retriever));
            }
            //当站点没启动的时候，这里可能出现问题
            string doc = await retriever.GetDocumentAsync(address, cancel).ConfigureAwait(false);

            LogHelper.LogVerbose(LogMessages.IDX21811, doc);
            OpenIdConnectConfiguration openIdConnectConfiguration = JsonConvert.DeserializeObject<OpenIdConnectConfiguration>(doc);
            if (!string.IsNullOrEmpty(openIdConnectConfiguration.JwksUri))
            {
                LogHelper.LogVerbose(LogMessages.IDX21812, openIdConnectConfiguration.JwksUri);
                //当证书配置或读取错误的时候，这里会出现问题
                string keys = await retriever.GetDocumentAsync(openIdConnectConfiguration.JwksUri, cancel).ConfigureAwait(false);

                LogHelper.LogVerbose(LogMessages.IDX21813, openIdConnectConfiguration.JwksUri);
                openIdConnectConfiguration.JsonWebKeySet = JsonConvert.DeserializeObject<JsonWebKeySet>(keys);
                foreach (SecurityKey key in openIdConnectConfiguration.JsonWebKeySet.GetSigningKeys())
                {
                    openIdConnectConfiguration.SigningKeys.Add(key);
                }
            }

            return openIdConnectConfiguration;
        }

通过流程进行问题解析：

通过查看日志可发现在流程5后开始报错

步骤一

访问：http://{ssoHost}/.well-known/openid-configuration

如果访问成功，则排除单点登录站点部署失败问题，前往步骤二

如果访问失败，则去解决单点登录站点部署问题。

步骤二

http://{ssoHost}/.well-known/openid-configuration/jwks

如果访问成功，则重启客户端

如果访问失败，那么有意思了...继续往下看

此时单点登录站点的日志

[10:25:24 Error] Microsoft.AspNetCore.Diagnostics.ExceptionHandlerMiddleware
An unhandled exception has occurred while executing the request.
Internal.Cryptography.CryptoThrowHelper+WindowsCryptographicException: 出现了内部错误。
   at Internal.Cryptography.Helpers.OpenStorageProvider(CngProvider provider)
   at System.Security.Cryptography.CngKey.Import(Byte[] keyBlob, String curveName, CngKeyBlobFormat format, CngProvider provider)
   at System.Security.Cryptography.CngKey.Import(Byte[] keyBlob, CngKeyBlobFormat format)
   at Internal.Cryptography.Pal.X509Pal.DecodePublicKey(Oid oid, Byte[] encodedKeyValue, Byte[] encodedParameters, ICertificatePal certificatePal)
   at System.Security.Cryptography.X509Certificates.PublicKey.get_Key()
   at Microsoft.IdentityModel.Tokens.X509SecurityKey.get_PublicKey()
   at IdentityServer4.ResponseHandling.DiscoveryResponseGenerator.CreateJwkDocumentAsync()
   at IdentityServer4.Endpoints.DiscoveryKeyEndpoint.ProcessAsync(HttpContext context)
   at IdentityServer4.Hosting.IdentityServerMiddleware.Invoke(HttpContext context, IEndpointRouter router, IUserSession session, IEventService events)
   at IdentityServer4.Hosting.IdentityServerMiddleware.Invoke(HttpContext context, IEndpointRouter router, IUserSession session, IEventService events)
   at Microsoft.AspNetCore.Authentication.AuthenticationMiddleware.Invoke(HttpContext context)
   at IdentityServer4.Hosting.BaseUrlMiddleware.Invoke(HttpContext context)
   at Microsoft.AspNetCore.StaticFiles.StaticFileMiddleware.Invoke(HttpContext context)
   at Microsoft.AspNetCore.Diagnostics.ExceptionHandlerMiddleware.Invoke(HttpContext context)

单点登录站点在此时的工作流程

graph TB

A(通过Http获取发现服务的配置信息)--1.进入SSO站点-->B(IdentityServerMiddleware)
B--2.调用Process-->C(进入DiscoveryKeyEndpoint)
C--3.遍历注入的加密Key-->D(返回JsonBWebKey)

部分源码解析：

   public async Task Invoke(HttpContext context, IEndpointRouter router, IUserSession session, IEventService events)
        {
            // this will check the authentication session and from it emit the check session
            // cookie needed from JS-based signout clients.
            await session.EnsureSessionIdCookieAsync();

            try
            {
                //根据请求路径，进入相应endpoint处理器
                var endpoint = router.Find(context);
                if (endpoint != null)
                {
                    _logger.LogInformation("Invoking IdentityServer endpoint: {endpointType} for {url}", endpoint.GetType().FullName, context.Request.Path.ToString());

                    var result = await endpoint.ProcessAsync(context);

                ...

  //创建JwtDocument
  public virtual async Task<IEnumerable<Models.JsonWebKey>> CreateJwkDocumentAsync()
        {
            var webKeys = new List<Models.JsonWebKey>();
            var signingCredentials = await Keys.GetSigningCredentialsAsync();
            var algorithm = signingCredentials?.Algorithm ?? Constants.SigningAlgorithms.RSA_SHA_256;

            foreach (var key in await Keys.GetValidationKeysAsync())
            {
                if (key is X509SecurityKey x509Key)
                {
                    var cert64 = Convert.ToBase64String(x509Key.Certificate.RawData);
                    var thumbprint = Base64Url.Encode(x509Key.Certificate.GetCertHash());
                    //当证书读取不正常的时候，PublicKey的属性构造器中会报错。
                    var pubKey = x509Key.PublicKey as RSA;
                    var parameters = pubKey.ExportParameters(false);
                    var exponent = Base64Url.Encode(parameters.Exponent);
                    var modulus = Base64Url.Encode(parameters.Modulus);

                    var webKey = new Models.JsonWebKey
                    {
                        kty = "RSA",
                        use = "sig",
                        kid = x509Key.KeyId,
                        x5t = thumbprint,
                        e = exponent,
                        n = modulus,
                        x5c = new[] { cert64 },
                        alg = algorithm
                    };

                    webKeys.Add(webKey);
                    continue;
                }

问题解析

通过日志可发现，问题出现在PublicKey的获取上，即证书的读取失败了。

解决方案！

终于到解决问题的时候了...

步骤一：

检查证书是否是临时证书

//在正式环境中，这可能会报错
builder.AddDeveloperSigningCredential(true, "tempkey.rsa");

//正确方式
builder.AddSigningCredential(new X509Certificate2(path,
                         Configuration["Certificates:Password"]))
这里可以参见郭的随笔：
https://www.cnblogs.com/guolianyu/p/9872661.html

步骤二：

如果已经添加了证书，且证书路径正确读取，且证书密码正确；

但是！在Windows Server2008服务器上部署仍失败时，请检查 CNG Key Isolation 服务是否启动！

如果 该服务未启动，则启动后即可！

如果该服务已启动，则尴尬的很...如果您仍排除不了问题，可以评论告诉我