KeyNotFoundException on /hangfire/succeeded page

[oliverjanik]

Normal nuget install. Default configuration.

[odinserj]

@oliverjanik, thank you for reporting this. Looks like there is need for additional check, but the case is very strange and I don't understand the cause.

Have you tried to refresh the page? Does the problem persist or disappear?

[oliverjanik]

It persists. I just did one console.writeline backgroundjob. Anything else I can give you?

[odinserj]

What version of Hangfire.Core and Hangfire.SqlServer (from packages.config) are you using? Can you show me the Hangfire.State table (and other Hangfire tables if possible)? (here or email odinserj gmail com)

[odinserj]

@oliverjanik, any news regarding to the problem problem?

[oliverjanik]

Sorry I won't have access to that computer until Tue or so. I'll try to replicate on my home machine.

[oliverjanik]

Here you go, replicated on different computer.

Versions:

<package id="Hangfire" version="1.3.1" targetFramework="net45" />
<package id="Hangfire.Core" version="1.3.1" targetFramework="net45" />
<package id="Hangfire.SqlServer" version="1.3.1" targetFramework="net45" />

Scripted Hangfire State:

INSERT INTO [HangFire].[State] ([Id], [JobId], [Name], [Reason], [CreatedAt], [Data]) VALUES (1, 1, N'Enqueued', NULL, N'2015-01-20 01:26:10', N'{"enqueuedAt":"2015-01-20T01:26:10.2286288Z","queue":"default"}')
INSERT INTO [HangFire].[State] ([Id], [JobId], [Name], [Reason], [CreatedAt], [Data]) VALUES (2, 1, N'Processing', NULL, N'2015-01-20 01:26:13', N'{"startedAt":"2015-01-20T01:26:13.6696288Z","serverId":"fr-130010007666:3548","workerNumber":"18"}')
INSERT INTO [HangFire].[State] ([Id], [JobId], [Name], [Reason], [CreatedAt], [Data]) VALUES (3, 1, N'Succeeded', NULL, N'2015-01-20 01:26:13', N'{"succeededAt":"2015-01-20T01:26:13.7936288Z","performanceDuration":"14","latency":"3538"}')

JobParameter (might be of interest):

INSERT INTO [HangFire].[JobParameter] ([Id], [JobId], [Name], [Value]) VALUES (1, 1, N'CurrentCulture', N'"en-AU"')
INSERT INTO [HangFire].[JobParameter] ([Id], [JobId], [Name], [Value]) VALUES (2, 1, N'CurrentUICulture', N'"en-US"')

[MichaelGG]

I'm hitting a similar bug too on succeeded, except I'm getting a NullReferenceException. Here's the top of the stack trace.

[NullReferenceException: Object reference not set to an instance of an object.]
   Hangfire.SqlServer.<>c__DisplayClass23.<SucceededJobs>b__22(SqlJob sqlJob, Job job, Dictionary`2 stateData) +142
   Hangfire.SqlServer.SqlServerMonitoringApi.DeserializeJobs(ICollection`1 jobs, Func`4 selector) +312
   Hangfire.SqlServer.SqlServerMonitoringApi.GetJobs(SqlConnection connection, Int32 from, Int32 count, String stateName, Func`4 selector) +254
   Hangfire.SqlServer.<>c__DisplayClass23.<SucceededJobs>b__21(SqlConnection connection) +232
   Hangfire.SqlServer.SqlServerMonitoringApi.UseConnection(Func`2 action) +204
   Hangfire.SqlServer.SqlServerMonitoringApi.SucceededJobs(Int32 from, Int32 count) +147
   Hangfire.Dashboard.Pages.SucceededJobs.Execute() +593
   Hangfire.Dashboard.RazorPage.TransformText(String innerContent) +68
   Hangfire.Dashboard.RazorPage.TransformText() +38
   Hangfire.Dashboard.RazorPageDispatcher.Dispatch(RequestDispatcherContext context) +297
   Hangfire.Dashboard.DashboardMiddleware.Invoke(IOwinContext context) +517

[odinserj]

@oliverjanik, thank you for the detailed response! Hangfire does not expect that property names are being serialized using camel-case convention, but it looks like it was set globally in your project. I'll find all the places that expect non-camelcase data and change them in 1.3.2.

As for now, you can use workaround by using separate serializer setting for Hangfire:

var settings = new JsonSerializerSettings();
Hangfire.Common.JobHelper.SetSerializerSettings(settings);

@MichaelGG, does your exception appear constantly or it is a transient one?

[oliverjanik]

@odinserj thanx, that makes a lot of sense.

[MichaelGG]

@odinserj It was transient - and I haven't been able to get it to happen again. If it does I'll be sure to capture all the SQL tables in use.

[odinserj]

@MichaelGG, thank you for the quick response. If it is transient, then I know the reason – job's state was changed during the fetch – and will fix it in 1.3.2.

[odinserj]

@oliverjanik, this bug is fixed now in just released 1.3.2. @MichaelGG, your bug is not fixed. Can you show me State table that causes your exception in a different GitHub issue?

[MichaelGG]

@odinserj I have not been able to repro it, sorry for not grabbing the tables when it happened :.

[odinserj]

@MichaelGG, no worries, this is a non-critical bug