Create logs for Identity Server 4 - c#

I'm trying to figure out how to get more detailed logs out of Identity Server 4, especially if things go wrong. I have read their documentation here http://docs.identityserver.io/en/latest/topics/logging.html However, I can't seem to get it to work. For one I don't use serilog and I have the IS4 running on a server I don't have remote access to so console logging is not going to work for me.
As such, I've tried to inject my own custom logging library that I have with the following:
public class Temp : IDisposable
{
public void Dispose() { }
}
public class CustomLogger : ILogger
{
private readonly IDatabaseLoggerService _databaseLoggerService;
public CustomLogger(IDatabaseLoggerService databaseLoggerService)
{
_databaseLoggerService = databaseLoggerService;
}
public void Log<TState>(LogLevel logLevel, EventId eventId, TState state, Exception exception, Func<TState, Exception, string> formatter)
{
if(exception == null) return;
_databaseLoggerService.Fatal(new LogErrorDetails
{
Exception = exception,
Message = "LIS - " + formatter
});
}
public bool IsEnabled(LogLevel logLevel)
{
return true;
}
public IDisposable BeginScope<TState>(TState state)
{
return new Temp();
}
}
public class CustomLoggerProvider : ILoggerProvider
{
private readonly IDatabaseLoggerService _databaseLoggerService;
public CustomLoggerProvider(IDatabaseLoggerService databaseLoggerService)
{
_databaseLoggerService = databaseLoggerService;
}
public void Dispose() { }
public ILogger CreateLogger(string categoryName)
{
return new CustomLogger(_databaseLoggerService);
}
}
As you can see its fairly straightforward because at this point I don't need it polished, just need it working, and I'll work on the proper implementation of it later once I have the sample working.
Now to set this up I did it in the Startup:
public void Configure(IApplicationBuilder app, IHostingEnvironment env, ILoggerFactory loggerFactory)
{
//code emmited
var serviceProvider = app.ApplicationServices;
var dbLogger = serviceProvider.GetService<IDatabaseLoggerService>();
loggerFactory.AddProvider(new CustomLoggerProvider(dbLogger));
}
However, I don't see any logs about it, the only thing I saw was the occasional anti-forgery token error that now pops up and that is it:
at Microsoft.AspNetCore.Antiforgery.Internal.DefaultAntiforgeryTokenSerializer.Deserialize(String serializedToken) at Microsoft.AspNetCore.Antiforgery.Internal.DefaultAntiforgery.GetCookieTokenDoesNotThrow(HttpContext httpContext)
What I would like to see is if a token request was successful or if it failed and why, but I'm not seeing any of it.
How would I go to accomplish higher level of logging for Identity Server 4?

You should be able to add your logger right into the WebHostBuilder giving you probably most insightful logs.
You can start of by creating an extension method to keep things nice and clean. We just need to add a singleton of our logger factory to the DI.
public static IWebHostBuilder UseCustomLogger(this IWebHostBuilder builder, CustomLogger logger = null, bool dispose = false)
{
if (builder == null) throw new ArgumentNullException(nameof(builder));
builder.ConfigureServices(collection =>
collection.AddSingleton<ILoggerFactory>(services => new CustomLoggerFactory(logger, dispose)));
return builder;
}
Factory itself would be very simple.
public class CustomLoggerFactory : ILoggerFactory
{
private readonly CustomLoggerProvider _provider;
public CustomLoggerFactory(ILogger logger = null, bool dispose = false)
{
_provider = new CustomLoggerProvider(logger, dispose);
}
public void Dispose()
{
_provider.Dispose();
}
public Microsoft.Extensions.Logging.ILogger CreateLogger(string categoryName)
{
return _provider.CreateLogger(categoryName);
}
}
You can then add your logger in Program.cs:
WebHost.CreateDefaultBuilder(args)
.UseStartup<Startup>()
.UseCustomLogger();
Lastly, looking at your logger implementation - you will need to create instance of IDatabaseLoggerService in the factory so that the logger can be created. Similar approach can of course be used with the ASP.NET Core DI.

Related

Aspnetboilerplate ObjectMapper or Logger is null

I have a small assembly that utilises the following NuGet packages Abp, Abp.AspNetCore and Abp.AbpAutoMapper.
For reference I am using Abp 6.4.0
I have added the depends to the module, the main service is also below.
The constructor inject is valid for both IObjectMapper and Ilogger. Whereas the inherited DomainService and ApplicationService for ObjectMapper and Logger are always null / NullLogger.Instance or NullObjectMapper.Instance.
[DependsOn(typeof(AbpAutoMapperModule))]
public class ServiceModule:AbpModule {
// other methods and logic
public override void PostInitialize(){
var mainService = Configuration.Get<IMainService>();
Task.Run(() =>
{
mainService.Start();
});
base.PostInitialize();
}
}
public class MainService : ApplicationService, IMainService
{
public MainService(IObjectMapper mapper, ILogger logger)
{
if (Logger == NullLogger.Instance)
{
Debug.WriteLine("No logger configured.....");
}
if (ObjectMapper == NullObjectMapper.Instance)
{
Debug.WriteLine("No mapper configured.....");
}
}
}
// OR domain service
public class MainService : DomainService, IMainService
{
public MainService(IObjectMapper mapper, ILogger logger)
{
if (Logger == NullLogger.Instance)
{
Debug.WriteLine("No logger configured.....");
}
if (ObjectMapper == NullObjectMapper.Instance)
{
Debug.WriteLine("No mapper configured.....");
}
}
}
Any ideas on how this can be resolved?
Thanks
I think, u have to register your injections. Something like this:
public class ServiceModule:AbpModule {
// other methods and logic
public override void ConfigureServices(ServiceConfigurationContext context)
{
context.Services.AddScoped<IMainService>(sp => sp.GetRequiredService<MainService>());
}
public override void PostInitialize(){
var mainService = Configuration.Get<IMainService>();
Task.Run(() =>
{
mainService.Start();
});
base.PostInitialize();
}
}
DomainService and ApplicationService allow IObjectMapper and ILogger to be property-injected.
If MainService requires non-null object pattern instances to work properly, you can proactively assign your constructor-injected instances:
public MainService(IObjectMapper mapper, ILogger logger)
{
ObjectMapper = mapper;
Logger = logger;
}
References:
https://aspnetboilerplate.com/Pages/Documents/Dependency-Injection#property-injection-pattern
https://aspnetboilerplate.com/Pages/Documents/Logging#getting-the-logger

Test ILogger with FakeItEasy

I'm working on a fun project for myself to learn blazor, a lot of core concepts and just general practice.
I'm trying to implement logger extension for FakeItEasy to easier test Microsofts ILogger.
Based on https://dev.azure.com/BrideButler/Bride%20Butler%20Open%20Source/_git/FakeItEasy.DotnetCore.LoggerExtensions?path=%2FFakeItEasy.DotnetCore.LoggerExtensions%2FLoggerExtensions.cs but now trying to get it to work for .NET 5.
Now that FormattedLogValues is internal this code doesn't work anymore.
The goal is to have something like this in my test (and not use Moq)
_logger.VerifyLog(LogLevel.Information, "Get Weather Called").MustHaveHappenedOnceExactly();
or A.CallTo(() => _logger.Log(LogLevel.Information, "Get Weather Called").MustHaveHappenedOnceExactly(); It doesn't have to have syntactic sugar but I would like to get it to work with FakeItEasy instead of Moq.
Below is all the involved code. Ofcourse I see some difference in the first line Microsoft.Extensions.Logging.ILogger.Log1[System.Object]vsMicrosoft.Extensions.Logging.ILogger1[Server.Controllers.WeatherForecastController].Log1[Microsoft.Extensions.Logging.FormattedLogValues]` but I've no clue how to "fix" this since I dont fully understand what is going wrong.
My question is obviously, what should I do to fix it, but also curious what part i'm missing.
My error is the following:
Assertion failed for the following call:
Microsoft.Extensions.Logging.ILogger.Log`1[System.Object]
(
logLevel: Information,
eventId: <Ignored>,
state: <e => e.ToString().Contains(value(Server.Tests.Extensions.LoggerExtensions+<>c__DisplayClass2_0`1[Server.Controllers.WeatherForecastController]).message)>,
exception: <Ignored>,
formatter: <Ignored>
)
Expected to find it once exactly but didn't find it among the calls:
1: Microsoft.Extensions.Logging.ILogger`1[Server.Controllers.WeatherForecastController].Log`1[Microsoft.Extensions.Logging.FormattedLogValues]
(
logLevel: Information,
eventId: 0,
state: [[{OriginalFormat}, Get Weather Called]],
exception: NULL,
formatter: System.Func`3[Microsoft.Extensions.Logging.FormattedLogValues,System.Exception,System.String]
)
Tests:
(using xUnit, FakeItEasy, Fixture)
LoggerExtensions
public static class LoggerExtensions
{
public static void VerifyLogMustHaveHappened<T>(this ILogger<T> logger, LogLevel level, string message)
{
try
{
logger.VerifyLog(level, message)
.MustHaveHappened();
}
catch (Exception e)
{
throw new ExpectationException($"while verifying a call to log with message: \"{message}\"", e);
}
}
public static void VerifyLogMustNotHaveHappened<T>(this ILogger<T> logger, LogLevel level, string message)
{
try
{
logger.VerifyLog(level, message)
.MustNotHaveHappened();
}
catch (Exception e)
{
throw new ExpectationException($"while verifying a call to log with message: \"{message}\"", e);
}
}
public static IVoidArgumentValidationConfiguration VerifyLog<T>(this ILogger<T> logger, LogLevel level,
string message)
{
return A.CallTo(() => logger.Log(
level,
A<EventId>._,
A<object>.That.Matches(e => e.ToString().Contains(message)),
A<Exception>._,
A<Func<object, Exception, string>>._)
);
}
}
TestConstructor:
private readonly IFixture _fixture;
private readonly ILogger<WeatherForecastController> _logger;
public WeatherForecastControllerTests()
{
_fixture = new Fixture().Customize(new AutoFakeItEasyCustomization());
_logger = _fixture.Freeze<Fake<ILogger<WeatherForecastController>>>().FakedObject;
}
Test:
[Fact]
public void WeatherForecast_Get_Should_Log()
{
// Arrange
var weatherController = new WeatherForecastController(_logger);
// Act
weatherController.Get();
// Assert
_logger.VerifyLog(LogLevel.Information, "Get Weather Called").MustHaveHappenedOnceExactly();
}
Controller:
public class WeatherForecastController : ControllerBase
{
private readonly ILogger _logger;
public WeatherForecastController(ILogger<WeatherForecastController> logger)
{
_logger = logger;
}
[HttpGet]
public IEnumerable<WeatherForecast> Get()
{
// Left out other code then logging
_logger.Log(LogLevel.Information, "Get Weather Called");
}
}
Update:
Options found
Only test if logger was called with correct log level:
public static IVoidArgumentValidationConfiguration VerifyLog<T>(this ILogger<T> logger, LogLevel level, string message)
{
return A.CallTo(logger).Where(call => call.Method.Name == "Log" && call.GetArgument<LogLevel>(0) == level);
}
Switch to Moq:
https://stackoverflow.com/a/59182490/1112413
https://stackoverflow.com/a/56728528/1112413
After reading https://stackoverflow.com/users/4728685/pavel-anikhouski comment on the post. I "got" what was wrong. Went to countless topics and made a solution that "works" tho probably could use some better coding standards. Hope it helps anyone else!
Edit: Some explaining.
Found this github comment that is the base of the solution. Then found this github pr where they mentioned to change Microsoft.Extensions.Logging.FormattedLogValues to IReadOnlyList<KeyValuePair<string, object>>
This can be extracted from GetArgument(2). (Source: https://stackoverflow.com/a/63701968/1112413)
public static class LoggerExtensions
{
public static void VerifyLogMustHaveHappened<T>(this ILogger<T> logger, LogLevel level, string message)
{
try
{
logger.VerifyLog(level, message)
.MustHaveHappened();
}
catch (Exception e)
{
throw new ExpectationException($"while verifying a call to log with message: \"{message}\"", e);
}
}
public static void VerifyLogMustNotHaveHappened<T>(this ILogger<T> logger, LogLevel level, string message)
{
try
{
logger.VerifyLog(level, message)
.MustNotHaveHappened();
}
catch (Exception e)
{
throw new ExpectationException($"while verifying a call to log with message: \"{message}\"", e);
}
}
public static IVoidArgumentValidationConfiguration VerifyLog<T>(this ILogger<T> logger, LogLevel level, string message)
{
return A.CallTo(logger)
.Where(call => call.Method.Name == "Log"
&& call.GetArgument<LogLevel>(0) == level
&& CheckLogMessages(call.GetArgument<IReadOnlyList<KeyValuePair<string, object>>>(2), message));
}
private static bool CheckLogMessages(IReadOnlyList<KeyValuePair<string, object>> readOnlyLists, string message)
{
foreach(var kvp in readOnlyLists)
{
if (kvp.Value.ToString().Contains(message))
{
return true;
}
}
return false;
}
}
Based on the same comments of Github
I ended up with a solution like that:
public class MyClassWithLoggerTests
{
private readonly MyClassWithLogger _sut;
private readonly ILogger<MyClassWithLogger> _logger;
public MyClassWithLoggerTests()
{
this._logger = A.Fake<ILogger<MyClassWithLogger>>();
this._sut = new MyClassWithLogger(this._logger);
}
[Fact]
public ATestMethodTitle()
{
this._sut.MethodToTest();
A.CallTo(_logger).Where(
call => call.Method.Name == "Log" && call.GetArgument<LogLevel>(0) == LogLevel.Information) // Modify with your type of log
.MustHaveHappened(1, Times.Exactly); // Modify with the number of times log is called
}
}

Make one more dependency injection inside already dependency injection class

I write console app on .net core and want to add Nlog into classes. I add dependency injection in program class
public static async Task Main(string[] args)
{
var logger = LogManager.GetCurrentClassLogger();
try
{
var configuration = GetConfiguration("app.config");
var serviceProvider = BuildDI(configuration);
using (serviceProvider as IDisposable)
{
var tcpController = serviceProvider.GetRequiredService<TcpController>();
var median = await tcpController.GetMedian();
Console.WriteLine(median.ToString());
}
}
catch (Exception ex)
{
logger.Error(ex);
throw;
}
finally
{
LogManager.Shutdown();
}
}
I generate ServiceProvider in this method
private static IServiceProvider BuildDI(IConfiguration config)
{
return new ServiceCollection()
.AddSingleton<IConfiguration>(config)
.AddSingleton<TcpController>()
.AddLogging(loggingBuilder =>
{
loggingBuilder.ClearProviders();
loggingBuilder.SetMinimumLevel(Microsoft.Extensions.Logging.LogLevel.Trace);
loggingBuilder.AddNLog(config);
})
.BuildServiceProvider();
}
In private static IConfiguration GetConfiguration(string configName) all logic to generate configuration object. I guess we don't need this code for this question.
In TcpController i got private TcpRepository _tcpRepository { get; } field that create in constructor.
public TcpController(IConfiguration configuration, ILogger<TcpController> logger)
{
this._configuration = configuration;
this._logger = logger;
this._tcpRepository = new TcpRepository(configuration);
}
I want to use it field like Transient. I must in TcpController repeat method private static IServiceProvider BuildDI(IConfiguration config) or in c# we have another way to make this?

Using Hub from separate class in ASP.NET Core 3

I'm working on a program where I receive data from SignalR, perform processing, and then send a SignalR message back to the client once the processing has finished. I've found a couple of resources for how to do this, but I can't quite figure out how to implement it in my project.
Here's what my code looks like:
Bootstrapping
public static void Main(string[] args)
{
CancellationTokenSource cancellationTokenSource = new CancellationTokenSource();
List<ISystem> systems = new List<ISystem>
{
new FirstProcessingSystem(),
new SecondProcessingSystem(),
};
Processor processor = new Processor(
cancellationToken: cancellationTokenSource.Token,
systems: systems);
processor.Start();
CreateHostBuilder(args).Build().Run();
cancellationTokenSource.Cancel();
}
public static IHostBuilder CreateHostBuilder(string[] args) =>
Host.CreateDefaultBuilder(args)
.ConfigureWebHostDefaults(webBuilder =>
{
webBuilder.UseStartup<Startup>();
});
public class Startup
{
// This method gets called by the runtime. Use this method to add services to the container.
// For more information on how to configure your application, visit https://go.microsoft.com/fwlink/?LinkID=398940
public void ConfigureServices(IServiceCollection services)
{
services.AddSignalR();
}
// This method gets called by the runtime. Use this method to configure the HTTP request pipeline.
public void Configure(IApplicationBuilder app, IWebHostEnvironment env)
{
if (env.IsDevelopment())
{
app.UseDeveloperExceptionPage();
}
app.UseRouting();
app.UseEndpoints(endpoints =>
{
endpoints.MapHub<TestHub>("/testHub");
});
}
}
TestHub.cs
public class TestHub : Hub
{
public async Task DoStuff(Work work)
{
FirstProcessingSystem.ItemsToProcess.Add(work);
}
}
Work.cs
public class Work
{
public readonly string ConnectionId;
public readonly string Data;
public Work(string connectionId, string data)
{
ConnectionId = connectionId;
Data = data;
}
}
Processor.cs
public class Processor
{
readonly CancellationToken CancellationToken;
readonly List<ISystem> Systems;
public Processor(
CancellationToken cancellationToken,
List<ISystem> systems)
{
CancellationToken = cancellationToken;
Systems = systems;
}
public void Start()
{
Task.Run(() =>
{
while (!CancellationToken.IsCancellationRequested)
{
foreach (var s in Systems)
s.Process();
}
});
}
}
Systems
public interface ISystem
{
void Process();
}
public class FirstProcessingSystem : ISystem
{
public static ConcurrentBag<Work> ItemsToProcess = new ConcurrentBag<Work>();
public void Process()
{
while (!ItemsToProcess.IsEmpty)
{
Work work;
if (ItemsToProcess.TryTake(out work))
{
// Do things...
SecondProcessingSystem.ItemsToProcess.Add(work);
}
}
}
}
public class SecondProcessingSystem : ISystem
{
public static ConcurrentBag<Work> ItemsToProcess = new ConcurrentBag<Work>();
public void Process()
{
while (!ItemsToProcess.IsEmpty)
{
Work work;
if (ItemsToProcess.TryTake(out work))
{
// Do more things...
// Hub.Send(work.ConnectionId, "Finished");
}
}
}
}
I know that I can perform the processing in the Hub and then send back the "Finished" call, but I'd like to decouple my processing from my inbound messaging that way I can add more ISystems when needed.
Can someone please with this? (Also, if someone has a better way to structure my program I'd also appreciate the feedback)
aspnet has a very powerful dependency injection system, why don't you use it? By creating your worker services without dependency injection, you'll have a hard time using anything provided by aspnet.
Since your "processing systems" seem to be long running services, you'd typically have them implement IHostedService, then create a generic service starter (taken from here):
public class BackgroundServiceStarter<T> : IHostedService where T : IHostedService
{
readonly T _backgroundService;
public BackgroundServiceStarter(T backgroundService)
{
_backgroundService = backgroundService;
}
public Task StartAsync(CancellationToken cancellationToken)
{
return _backgroundService.StartAsync(cancellationToken);
}
public Task StopAsync(CancellationToken cancellationToken)
{
return _backgroundService.StopAsync(cancellationToken);
}
}
then register them to the DI container in ConfigureServices:
// make the classes injectable
services.AddSingleton<FirstProcessingSystem>();
services.AddSingleton<SecondProcessingSystem>();
// start them up
services.AddHostedService<BackgroundServiceStarter<FirstProcessingSystem>>();
services.AddHostedService<BackgroundServiceStarter<SecondProcessingSystem>>();
Now that you got all that set up correctly, you can simply inject a reference to your signalR hub using IHubContext<TestHub> context in the constructor parameters of whatever class that needs it (as described in some of the links you posted).

Hangfire dependency injection with .NET Core

How can I use .NET Core's default dependency injection in Hangfire?
I am new to Hangfire and searching for an example which works with ASP.NET Core.
See full example on GitHub https://github.com/gonzigonz/HangfireCore-Example.
Live site at http://hangfirecore.azurewebsites.net/
Make sure you have the Core version of Hangfire:
dotnet add package Hangfire.AspNetCore
Configure your IoC by defining a JobActivator. Below is the config for use with the default asp.net core container service:
public class HangfireActivator : Hangfire.JobActivator
{
private readonly IServiceProvider _serviceProvider;
public HangfireActivator(IServiceProvider serviceProvider)
{
_serviceProvider = serviceProvider;
}
public override object ActivateJob(Type type)
{
return _serviceProvider.GetService(type);
}
}
Next register hangfire as a service in the Startup.ConfigureServices method:
services.AddHangfire(opt =>
opt.UseSqlServerStorage("Your Hangfire Connection string"));
Configure hangfire in the Startup.Configure method. In relationship to your question, the key is to configure hangfire to use the new HangfireActivator we just defined above. To do so you will have to provide hangfire with the IServiceProvider and this can be achieved by just adding it to the list of parameters for the Configure method. At runtime, DI will providing this service for you:
public void Configure(
IApplicationBuilder app,
IHostingEnvironment env,
ILoggerFactory loggerFactory,
IServiceProvider serviceProvider)
{
...
// Configure hangfire to use the new JobActivator we defined.
GlobalConfiguration.Configuration
.UseActivator(new HangfireActivator(serviceProvider));
// The rest of the hangfire config as usual.
app.UseHangfireServer();
app.UseHangfireDashboard();
}
When you enqueue a job, use the registered type which usually is your interface. Don't use a concrete type unless you registered it that way. You must use the type registered with your IoC else Hangfire won't find it.
For Example say you've registered the following services:
services.AddScoped<DbManager>();
services.AddScoped<IMyService, MyService>();
Then you could enqueue DbManager with an instantiated version of the class:
BackgroundJob.Enqueue(() => dbManager.DoSomething());
However you could not do the same with MyService. Enqueuing with an instantiated version would fail because DI would fail as only the interface is registered. In this case you would enqueue like this:
BackgroundJob.Enqueue<IMyService>( ms => ms.DoSomething());
DoritoBandito's answer is incomplete or deprecated.
public class EmailSender {
public EmailSender(IDbContext dbContext, IEmailService emailService)
{
_dbContext = dbContext;
_emailService = emailService;
}
}
Register services:
services.AddTransient<IDbContext, TestDbContext>();
services.AddTransient<IEmailService, EmailService>();
Enqueue:
BackgroundJob.Enqueue<EmailSender>(x => x.Send(13, "Hello!"));
Source:
http://docs.hangfire.io/en/latest/background-methods/passing-dependencies.html
Note: if you want a full sample, see my blog post on this.
All of the answers in this thread are wrong/incomplete/outdated. Here's an example with ASP.NET Core 3.1 and Hangfire.AspnetCore 1.7.
Client:
//...
using Hangfire;
// ...
public class Startup
{
// ...
public void ConfigureServices(IServiceCollection services)
{
//...
services.AddHangfire(config =>
{
// configure hangfire per your requirements
});
}
}
public class SomeController : ControllerBase
{
private readonly IBackgroundJobClient _backgroundJobClient;
public SomeController(IBackgroundJobClient backgroundJobClient)
{
_backgroundJobClient = backgroundJobClient;
}
[HttpPost("some-route")]
public IActionResult Schedule([FromBody] SomeModel model)
{
_backgroundJobClient.Schedule<SomeClass>(s => s.Execute(model));
}
}
Server (same or different application):
{
//...
services.AddScoped<ISomeDependency, SomeDependency>();
services.AddHangfire(hangfireConfiguration =>
{
// configure hangfire with the same backing storage as your client
});
services.AddHangfireServer();
}
public interface ISomeDependency { }
public class SomeDependency : ISomeDependency { }
public class SomeClass
{
private readonly ISomeDependency _someDependency;
public SomeClass(ISomeDependency someDependency)
{
_someDependency = someDependency;
}
// the function scheduled in SomeController
public void Execute(SomeModel someModel)
{
}
}
As far as I am aware, you can use .net cores dependency injection the same as you would for any other service.
You can use a service which contains the jobs to be executed, which can be executed like so
var jobId = BackgroundJob.Enqueue(x => x.SomeTask(passParamIfYouWish));
Here is an example of the Job Service class
public class JobService : IJobService
{
private IClientService _clientService;
private INodeServices _nodeServices;
//Constructor
public JobService(IClientService clientService, INodeServices nodeServices)
{
_clientService = clientService;
_nodeServices = nodeServices;
}
//Some task to execute
public async Task SomeTask(Guid subject)
{
// Do some job here
Client client = _clientService.FindUserBySubject(subject);
}
}
And in your projects Startup.cs you can add a dependency as normal
services.AddTransient< IClientService, ClientService>();
Not sure this answers your question or not
Currently, Hangfire is deeply integrated with Asp.Net Core. Install Hangfire.AspNetCore to set up the dashboard and DI integration automatically. Then, you just need to define your dependencies using ASP.NET core as always.
If you are trying to quickly set up Hangfire with ASP.NET Core (tested in ASP.NET Core 2.2) you can also use Hangfire.MemoryStorage. All the configuration can be performed in Startup.cs:
using Hangfire;
using Hangfire.MemoryStorage;
public void ConfigureServices(IServiceCollection services)
{
services.AddHangfire(opt => opt.UseMemoryStorage());
JobStorage.Current = new MemoryStorage();
}
protected void StartHangFireJobs(IApplicationBuilder app, IServiceProvider serviceProvider)
{
app.UseHangfireServer();
app.UseHangfireDashboard();
//TODO: move cron expressions to appsettings.json
RecurringJob.AddOrUpdate<SomeJobService>(
x => x.DoWork(),
"* * * * *");
RecurringJob.AddOrUpdate<OtherJobService>(
x => x.DoWork(),
"0 */2 * * *");
}
public void Configure(IApplicationBuilder app, IServiceProvider serviceProvider)
{
StartHangFireJobs(app, serviceProvider)
}
Of course, everything is store in memory and it is lost once the application pool is recycled, but it is a quick way to see that everything works as expected with minimal configuration.
To switch to SQL Server database persistence, you should install Hangfire.SqlServer package and simply configure it instead of the memory storage:
services.AddHangfire(opt => opt.UseSqlServerStorage(Configuration.GetConnectionString("Default")));
I had to start HangFire in main function. This is how I solved it:
public static void Main(string[] args)
{
var host = CreateWebHostBuilder(args).Build();
using (var serviceScope = host.Services.CreateScope())
{
var services = serviceScope.ServiceProvider;
try
{
var liveDataHelper = services.GetRequiredService<ILiveDataHelper>();
var justInitHangfire = services.GetRequiredService<IBackgroundJobClient>();
//This was causing an exception (HangFire is not initialized)
RecurringJob.AddOrUpdate(() => liveDataHelper.RePopulateAllConfigDataAsync(), Cron.Daily());
// Use the context here
}
catch (Exception ex)
{
var logger = services.GetRequiredService<ILogger<Program>>();
logger.LogError(ex, "Can't start " + nameof(LiveDataHelper));
}
}
host.Run();
}
public static IWebHostBuilder CreateWebHostBuilder(string[] args) =>
WebHost.CreateDefaultBuilder(args)
.UseStartup<Startup>();
}
Actually there is an easy way for dependency injection based job registration.
You just need to use the following code in your Startup:
public class Startup {
public void Configure(IApplicationBuilder app)
{
var factory = app.ApplicationServices
.GetService<IServiceScopeFactory>();
GlobalConfiguration.Configuration.UseActivator(
new Hangfire.AspNetCore.AspNetCoreJobActivator(factory));
}
}
However i personally wanted a job self registration including on demand jobs (recurring jobs which are never executed, except by manual trigger on hangfire dashboard), which was a little more complex then just that. I was (for example) facing issues with the job service activation, which is why i decided to share most of my implementation code.
//I wanted an interface to declare my jobs, including the job Id.
public interface IBackgroundJob {
string Id { get; set; }
void Invoke();
}
//I wanted to retrieve the jobs by id. Heres my extension method for that:
public static IBackgroundJob GetJob(
this IServiceProvider provider,
string jobId) => provider
.GetServices<IBackgroundJob>()
.SingleOrDefault(j => j.Id == jobId);
//Now i needed an invoker for these jobs.
//The invoker is basically an example of a dependency injected hangfire job.
internal class JobInvoker {
public JobInvoker(IServiceScopeFactory factory) {
Factory = factory;
}
public IServiceScopeFactory Factory { get; }
public void Invoke(string jobId)
{
//hangfire jobs should always be executed within their own scope.
//The default AspNetCoreJobActivator should technically already do that.
//Lets just say i have trust issues.
using (var scope = Factory.CreateScope())
{
scope.ServiceProvider
.GetJob(jobId)?
.Invoke();
}
}
//Now i needed to tell hangfire to use these jobs.
//Reminder: The serviceProvider is in IApplicationBuilder.ApplicationServices
public static void RegisterJobs(IServiceProvider serviceProvider) {
var factory = serviceProvider.GetService();
GlobalConfiguration.Configuration.UseActivator(new Hangfire.AspNetCore.AspNetCoreJobActivator(factory));
var manager = serviceProvider.GetService<IRecurringJobManager>();
var config = serviceProvider.GetService<IConfiguration>();
var jobs = serviceProvider.GetServices<IBackgroundJob>();
foreach (var job in jobs) {
var jobConfig = config.GetJobConfig(job.Id);
var schedule = jobConfig?.Schedule; //this is a cron expression
if (String.IsNullOrWhiteSpace(schedule))
schedule = Cron.Never(); //this is an on demand job only!
manager.AddOrUpdate(
recurringJobId: job.Id,
job: GetJob(job.Id),
cronExpression: schedule);
}
//and last but not least...
//My Method for creating the hangfire job with injected job id
private static Job GetJob(string jobId)
{
var type = typeof(JobInvoker);
var method = type.GetMethod("Invoke");
return new Job(
type: type,
method: method,
args: jobId);
}
Using the above code i was able to create hangfire job services with full dependency injection support. Hope it helps someone.
Use the below code for Hangfire configuration
using eForms.Core;
using Hangfire;
using Hangfire.SqlServer;
using System;
using System.ComponentModel;
using System.Web.Hosting;
namespace eForms.AdminPanel.Jobs
{
public class JobManager : IJobManager, IRegisteredObject
{
public static readonly JobManager Instance = new JobManager();
//private static readonly TimeSpan ZeroTimespan = new TimeSpan(0, 0, 10);
private static readonly object _lockObject = new Object();
private bool _started;
private BackgroundJobServer _backgroundJobServer;
private JobManager()
{
}
public int Schedule(JobInfo whatToDo)
{
int result = 0;
if (!whatToDo.IsRecurring)
{
if (whatToDo.Delay == TimeSpan.Zero)
int.TryParse(BackgroundJob.Enqueue(() => Run(whatToDo.JobId, whatToDo.JobType.AssemblyQualifiedName)), out result);
else
int.TryParse(BackgroundJob.Schedule(() => Run(whatToDo.JobId, whatToDo.JobType.AssemblyQualifiedName), whatToDo.Delay), out result);
}
else
{
RecurringJob.AddOrUpdate(whatToDo.JobType.Name, () => RunRecurring(whatToDo.JobType.AssemblyQualifiedName), Cron.MinuteInterval(whatToDo.Delay.TotalMinutes.AsInt()));
}
return result;
}
[DisplayName("Id: {0}, Type: {1}")]
[HangFireYearlyExpirationTime]
public static void Run(int jobId, string jobType)
{
try
{
Type runnerType;
if (!jobType.ToType(out runnerType)) throw new Exception("Provided job has undefined type");
var runner = runnerType.CreateInstance<JobRunner>();
runner.Run(jobId);
}
catch (Exception ex)
{
throw new JobException($"Error while executing Job Id: {jobId}, Type: {jobType}", ex);
}
}
[DisplayName("{0}")]
[HangFireMinutelyExpirationTime]
public static void RunRecurring(string jobType)
{
try
{
Type runnerType;
if (!jobType.ToType(out runnerType)) throw new Exception("Provided job has undefined type");
var runner = runnerType.CreateInstance<JobRunner>();
runner.Run(0);
}
catch (Exception ex)
{
throw new JobException($"Error while executing Recurring Type: {jobType}", ex);
}
}
public void Start()
{
lock (_lockObject)
{
if (_started) return;
if (!AppConfigSettings.EnableHangFire) return;
_started = true;
HostingEnvironment.RegisterObject(this);
GlobalConfiguration.Configuration
.UseSqlServerStorage("SqlDbConnection", new SqlServerStorageOptions { PrepareSchemaIfNecessary = false })
//.UseFilter(new HangFireLogFailureAttribute())
.UseLog4NetLogProvider();
//Add infinity Expiration job filter
//GlobalJobFilters.Filters.Add(new HangFireProlongExpirationTimeAttribute());
//Hangfire comes with a retry policy that is automatically set to 10 retry and backs off over several mins
//We in the following remove this attribute and add our own custom one which adds significant backoff time
//custom logic to determine how much to back off and what to to in the case of fails
// The trick here is we can't just remove the filter as you'd expect using remove
// we first have to find it then save the Instance then remove it
try
{
object automaticRetryAttribute = null;
//Search hangfire automatic retry
foreach (var filter in GlobalJobFilters.Filters)
{
if (filter.Instance is Hangfire.AutomaticRetryAttribute)
{
// found it
automaticRetryAttribute = filter.Instance;
System.Diagnostics.Trace.TraceError("Found hangfire automatic retry");
}
}
//Remove default hangefire automaticRetryAttribute
if (automaticRetryAttribute != null)
GlobalJobFilters.Filters.Remove(automaticRetryAttribute);
//Add custom retry job filter
GlobalJobFilters.Filters.Add(new HangFireCustomAutoRetryJobFilterAttribute());
}
catch (Exception) { }
_backgroundJobServer = new BackgroundJobServer(new BackgroundJobServerOptions
{
HeartbeatInterval = new System.TimeSpan(0, 1, 0),
ServerCheckInterval = new System.TimeSpan(0, 1, 0),
SchedulePollingInterval = new System.TimeSpan(0, 1, 0)
});
}
}
public void Stop()
{
lock (_lockObject)
{
if (_backgroundJobServer != null)
{
_backgroundJobServer.Dispose();
}
HostingEnvironment.UnregisterObject(this);
}
}
void IRegisteredObject.Stop(bool immediate)
{
Stop();
}
}
}
Admin Job Manager
public class Global : System.Web.HttpApplication
{
void Application_Start(object sender, EventArgs e)
{
if (Core.AppConfigSettings.EnableHangFire)
{
JobManager.Instance.Start();
new SchedulePendingSmsNotifications().Schedule(new Core.JobInfo() { JobId = 0, JobType = typeof(SchedulePendingSmsNotifications), Delay = TimeSpan.FromMinutes(1), IsRecurring = true });
}
}
protected void Application_End(object sender, EventArgs e)
{
if (Core.AppConfigSettings.EnableHangFire)
{
JobManager.Instance.Stop();
}
}
}

Categories

Resources