Hangfire InMemory 0.9.0 Object disposed exception

[Cynicszm]

Hi,

We are now getting object disposed exceptions in NUnit (4.1.0) after upgrading Hangfire InMemory from 0.8.1 to 0.9.0

Individual tests pass but fail on committing the write transaction when run concurrently.

Code snippet: `public string CreateToken( TimeSpan expireIn ) { var tokenId = Guid.NewGuid().ToString( "N" ); var resourceKey = "execution-tokens:" + tokenId;

using var connection = JobStorage.Current.GetConnection();
using var transaction = connection.CreateWriteTransaction();
transaction.SetRangeInHash( resourceKey, new Dictionary<string, string>
{
    { "Cancelled", "false" },
    { "Paused", "false" }
} );

// if storage supports manual expiration handling
if ( transaction is JobStorageTransaction jsTransaction )
{
    jsTransaction.ExpireHash( resourceKey, expireIn );
}
transaction.Commit();

return tokenId;

}`

Stack: System.ObjectDisposedException : Cannot access a disposed object. Object name: 'Hangfire.InMemory.InMemoryDispatcher'. at Hangfire.InMemory.InMemoryDispatcher.ThrowObjectDisposedException() in //src/Hangfire.InMemory/InMemoryDispatcher.cs:line 136 at Hangfire.InMemory.InMemoryDispatcher.QueryAndWait(Func`3 query) in //src/Hangfire.InMemory/InMemoryDispatcher.cs:line 58 at Hangfire.InMemory.InMemoryDispatcherBase.QueryAndWait[T](Func3 query) in /_/src/Hangfire.InMemory/InMemoryDispatcherBase.cs:line 126 at Hangfire.InMemory.InMemoryDispatcherBase.QueryAndWait(Action2 query) in //src/Hangfire.InMemory/InMemoryDispatcherBase.cs:line 120 at Hangfire.InMemory.InMemoryTransaction.Commit() in //src/Hangfire.InMemory/InMemoryTransaction.cs:line 51

[odinserj]

InMemoryStorage class now implements IDisposable. So something is calling the Dispose method. Try creating instances of this class explicitly for each test, instead of relying on a static instance in JobStorage.Current. In this case, unit tests will be properly isolated, and you will not have any shared state between test runs.

[Cynicszm]

Thanks for the quick response.

I've moved some of the JobStorage usages around and it appears to be behaving for now.

Appreciate the help.

[odinserj]

Great, thanks for the update!