Null Reference in GetServerTransactionLevel

[ShaneCourtrille]

I'm investigating an exception occurring when our system is under load and wanted to know if anyone has any idea what could cause the following? From reading the code it looks like the _innerConnection.Owner must be null but not sure what would cause that?

System.NullReferenceException:
   at Microsoft.Data.SqlClient.SqlInternalTransaction.GetServerTransactionLevel (Microsoft.Data.SqlClient, Version=2.0.20168.4, Culture=neutral, PublicKeyToken=23ec7fc2d6eaa4a5)
   at Microsoft.Data.SqlClient.SqlInternalTransaction.CheckTransactionLevelAndZombie (Microsoft.Data.SqlClient, Version=2.0.20168.4, Culture=neutral, PublicKeyToken=23ec7fc2d6eaa4a5)
   at Microsoft.Data.SqlClient.SqlInternalTransaction.Commit (Microsoft.Data.SqlClient, Version=2.0.20168.4, Culture=neutral, PublicKeyToken=23ec7fc2d6eaa4a5)
   at Microsoft.Data.SqlClient.SqlTransaction.Commit (Microsoft.Data.SqlClient, Version=2.0.20168.4, Culture=neutral, PublicKeyToken=23ec7fc2d6eaa4a5)
   at YesSql.Session.CommitTransaction (YesSql.Core, Version=0.0.0.0, Culture=neutral, PublicKeyToken=51cdb8c1938e4789)
   at YesSql.Session+<CommitAsync>d__39.MoveNext (YesSql.Core, Version=0.0.0.0, Culture=neutral, PublicKeyToken=51cdb8c1938e4789)

[sebastienros]

_innerConnection.Owner points to this code:
https://referencesource.microsoft.com/#System.Data/fx/src/data/System/Data/ProviderBase/DbConnectionInternal.cs,294

Which is a WeakReference so I believe the getter should create a local copy of Target and check for nullability.

https://referencesource.microsoft.com/#System.Data/fx/src/data/System/Data/ProviderBase/DbConnectionInternal.cs,39

[sebastienros]

Could also show a race in the pooling logic since reports of this issue seem to relate to high load.

[Wraith2]

use of WeakReference should be converted to WeakReference<T> if possible because the generic version has a usage pattern designed to avoid this sort of problem with time of check time of use dsparity.

[ShaneCourtrille]

The current code also loses the original exception which from our point of view is the most important one.

[ShaneCourtrille]

FYI the problem went away after we 1) increased the Min threads in ThreadPool 2) fixed Redis timeouts.. I'm not yet sure if 2) was causing the db transaction to timeout/fail which might be the cause.