Early run resulting in duplicate execution of the job

[RShergold]

Describe the bug
Jobs are being run a fraction of a second early resulting in duplicate execution of the job.

I am creating a single job with

services.AddNCronJob(options =>
{
    var timeZone = TimeZoneInfo.FindSystemTimeZoneById("GMT Standard Time");
    options.AddJob<UnreadMessageEmailJob>(job =>
    {
        job.WithCronExpression("17 12 * * *", timeZone);
    });
});

and then

    var app = builder.Build();

    await app.UseNCronJobAsync();
    await app.RunAsync();

Current behavior
In my logs I can see the job is consumed a fraction of a second early meaning it schedules the next job for the exact same time as the current job.

Then after the first job has run...

This results in two duplicate jobs with different correlation IDs being run back to back

Expected behavior
For the job to be run on or after the time defined in the cron expression.

When removing and adding the next job is it possible to check the date time of the two to make sure we are not adding the same time we just removed?

The app is running on an IIS server. I'm not sure if there is some kind of strange configuration on our server causing this issue as I do not have direct access to it.

Version information

• NCronJobVersion: 4.7.0
• .NET runtime version: net8.0

[linkdotnet]

I really have a hard time reproducing that issue (aka I can't reproduce it).
At least with a minimal setup like:

using NCronJob;

var builder = Host.CreateApplicationBuilder(args);

builder.Services.AddLogging();
builder.Services.AddNCronJob(options =>
{
    options.AddJob<UnreadMessageEmailJob>(job =>
    {
        job.WithCronExpression("23 11 * * *", TimeZoneInfo.Local);
    });
});

var app = builder.Build();
await app.UseNCronJobAsync();
await app.RunAsync();

internal sealed class UnreadMessageEmailJob : IJob
{
    public Task RunAsync(IJobExecutionContext context, CancellationToken token)
    {
        Console.WriteLine("Hey GitHub");
        return Task.CompletedTask;
    }
}

Do you have any additional configuration? LIke RetryPolicyAttribute (which would perfectly explain the 1 second)?
So it might be that your job has thrown an exception, NCronJob picked it up ad retried it.
Could that be?

[RShergold]

Hey. I appreciate you taking the time to look into this. I also realize I haven't given you much to work with 😄

I strongly suspect this is something to do with a strange setup on our servers.

As things wind down here for Christmas I'm going to try and find some time to dig into this a little deeper and see if I can isolate the issue more and maybe create something minimal and reproducible on other hardware.

[RShergold]

Long story short I haven't been able to recreate it on a new fresh server. But I can recreate it on two different test servers we have.

I stripped everything back to the minimum I need to work with our CI/CD pipelines which leaves me with

using System;

using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;

using NCronJob;

using Serilog;

try
{
    var builder = Host.CreateDefaultBuilder(args)
        .UseWindowsService()
        .UseSerilog((context, loggerConfiguration) =>
            loggerConfiguration.ReadFrom.Configuration(context.Configuration));

    var lastExecutedTime = DateTimeOffset.MinValue;

    builder.ConfigureServices((context, services) =>
    {
        services.AddNCronJob(
            (ILogger<Program> logger, TimeProvider timeProvider) =>
        {
            logger.LogInformation("NCronJob triggered at {Time:HH:mm:ss.ffffff }", timeProvider.GetUtcNow());

            if ((timeProvider.GetUtcNow() - lastExecutedTime).TotalMinutes < 1)
            {
                logger.LogError("The job executed less than a minute ago.");
            }

            lastExecutedTime = timeProvider.GetUtcNow();
        }, "*/2 * * * *");
    });

    var app = builder.Build();

    Log.Information("Starting NCronJob");
    await app.UseNCronJobAsync();

    Log.Information("Starting App");
    await app.RunAsync();
}
catch (Exception ex)
{
    Log.Fatal(ex, "Application terminated unexpectedly");
}

This consistently double executes on two different test servers with the same spec. Our servers are Azure VMs. I tried spinning up a couple of new VMs however the spec we use is not available to me on my personal account. When running this code against fresh VMs using a Windows Server 2025 Datacenter Azure Edition image I could not recreate the issue.

I tried running the above as a console app and as a Windows Service. (Since this app runs as a service on our servers)

Our test servers (where I can reliably recreate the issue) have the following spec:

Operating System: Windows (Windows Server 2019 Datacenter)
Size: Standard DS11 v2 (2 vcpus, 14 Gib memory)

I'm not sure If I can but I will try and get my company to spin up a fresh Windows (Windows Server 2019 Datacenter) instance. That way I can see if it's something we're running on the server or if it's due to some defect in Windows Server 2019 & .Net 8

[linkdotnet]

Very very interesting and I am not sure what should cause that to be frank.

If you remove the await app.UseNCronJobAsync(); part - does it still execute the job twice?

Just to get a grasp what could be the core issue (aka scheduling vs startup phase).