Disposing a disposed object

[RichardChen820]

Describe the bug
My application is running in Kubernetes, when I do deployment the old pod raised following exception when it shuts down. In my code, it just call CosmosClinet.Dispose() in object Dispose method.

Unhandled exception. System.ObjectDisposedException: Cannot access a disposed object.
Object name: 'Channel'.
at Microsoft.Azure.Documents.Rntbd.Channel.System.IDisposable.Dispose()
at Microsoft.Azure.Documents.Rntbd.LbChannelState.Dispose()
at Microsoft.Azure.Documents.Rntbd.LoadBalancingPartition.Dispose()
at Microsoft.Azure.Documents.Rntbd.LoadBalancingChannel.System.IDisposable.Dispose()
at Microsoft.Azure.Documents.Rntbd.LoadBalancingChannel.Close()
at Microsoft.Azure.Documents.Rntbd.ChannelDictionary.Dispose()
at Microsoft.Azure.Documents.Rntbd.TransportClient.Dispose()
at Microsoft.Azure.Documents.StoreClientFactory.Dispose()
at Microsoft.Azure.Cosmos.DocumentClient.Dispose()
at Microsoft.Azure.Cosmos.ClientContextCore.Dispose(Boolean disposing)
at Microsoft.Azure.Cosmos.ClientContextCore.Dispose()
at Microsoft.Azure.Cosmos.CosmosClient.Dispose(Boolean disposing)
at Microsoft.Azure.Cosmos.CosmosClient.Dispose()
at Microsoft.AppConfig.Agent.Storage.CosmosDbStorage.Dispose() in C:\__w\1\s\src\Microsoft.AppConfig.Agent.Storage\CosmosDbStorage.cs:line 47
   at Microsoft.Extensions.DependencyInjection.ServiceLookup.ServiceProviderEngineScope.DisposeAsync()
--- End of stack trace from previous location ---
at Microsoft.Extensions.Hosting.Internal.Host.DisposeAsync()
at Microsoft.Extensions.Hosting.Internal.Host.Dispose()
at Microsoft.AppConfig.Agent.Program.Main(String[] args) in C:\__w\1\s\src\Microsoft.AppConfig.Agent\Program.cs:line 35
at Microsoft.AppConfig.Agent.Program.<Main>(String[] args)

Environment summary
SDK Version: 3.43.0
OS Version: linux

[ananth7592]

Hi @RichardChen820 , I would appreciate some clarification on a few points to help me understand the issue better:

• Is the CosmosClient being utilized as a singleton instance throughout the application lifecycle within the pod?
• Does this exception occur every time a K8s deployment is performed, during the shutdown of the pod?
• Is this a regular deployment? If possible, could you please provide the repro steps to help me replicate the issue?

[RichardChen820]

@ananth7592 Thanks for helping look into this
Is the CosmosClient being utilized as a singleton instance throughout the application lifecycle within the pod?

Yes, CosmosClient is used as a singleton in my app.

Does this exception occur every time a K8s deployment is performed, during the shutdown of the pod?

No, it only happened once, I haven't seen it again after the last occurrence.

Is this a regular deployment? If possible, could you please provide the repro steps to help me replicate the issue?

This is a regular deployment that is performed nearly every two weeks. As of now we can't reproduce it as well