SslStream ObjectDisposedException causes reconnection loop in .NET 10 with ASP.Net 10/3.16.0 client

[chriscameron-vertexinc]

Describe the bug
We are experiencing a reconnection loop due to an unhandled System.ObjectDisposedException on SslStream in our application, using the latest Pulsar.Client (3.16.0), .NET 10, and official ASP.NET 10 Docker base images. This appears to leave the application stuck in a reconnection cycle instead of recovering or failing gracefully.

Exception details

System.ObjectDisposedException: Cannot access a disposed object.
Object name: 'System.Net.Security.SslStream'.
  at System.Net.Security.SslStream.<ThrowIfExceptional>g__ThrowExceptional|131_0(ExceptionDispatchInfo e)
  at Pipelines.Sockets.Unofficial.StreamConnection.AsyncStreamPipe.CopyFromWritePipeToStream()
  at System.IO.Pipelines.PipeWriter.CopyFromAsync(Stream source, CancellationToken cancellationToken)
  at Pulsar.Client.Common.Commands.serializeSimpleCommand@44-1.MoveNext()
  at <StartupCode$Pulsar-Client>.$ClientCnx.clo@302-10.MoveNext()

Environment:

• .NET 10
• Pulsar.Client 3.16.0
• Official ASP.NET 10 Docker image

Observed behavior

• The connection drops, triggering a reconnect loop.
• The reconnect logic appears to run repeatedly (see ConnectionHandler.fs, ConnectionPool.fs, ClientCnx.fs), but we never recover nor get a permanent failure.
• Exception is thrown in the context of SslStream/Pipelines, possibly after dispose.

Relevant code (semantic search references):

• src/Pulsar.Client/Internal/ConnectionPool.fs:142-155 – SslStream construction
• src/Pulsar.Client/Internal/ConnectionHandler.fs:64-83 – ReconnectLoop
• src/Pulsar.Client/Internal/ClientCnx.fs:805-848 – Socket read/reconnect handling

Possible cause

• The reconnect logic with SslStream/Sockets appears to trigger after an ObjectDisposedException that isn't fully handled, leading to continuous attempts.
• See dotnet/runtime #103499 for background around SslStream behavior on disconnection.

Expected behavior

• Client should either recover, or escalate to a permanent failure instead of looping on reconnect when SslStream is disposed.

Steps to reproduce

1. Deploy an application using ASP.NET 10 and Pulsar.Client 3.16.0.
2. Disrupt broker connection so SslStream gets disposed (e.g., network drop or broker shutdown).
3. Observe client logs for reconnect loop with ObjectDisposedException.

[Lanayx]

@chriscameron-vertexinc From the logs you attached, I see the following errors repeated:

{
  "MessageTemplate": "{0} Close connection because received internal-server error {1}",
  "Level": "Error",
  "Properties": {
    "0": "clientCnx(68, LogicalAddres Unspecified/vespa-pulsar-dev-proxy.platdev.vtxdev.net:6651)",
    "1": "{\"errorMsg\":\"\",\"reqId\":662259599059308581, \"remote\":\"vespa-pulsar-dev-us-east-2-broker-headless.vespa-pulsar-cluster-dev.svc.cluster.local/100.73.84.96:6650\", \"local\":\"/100.73.81.120:36724\"}",
    "AssemblyVersion": "1.12.0+92c6a7b2c89511b7782fe573605997fe3afb97d5",
    "dd_service": "vitr-reports",
    "AssemblyBuildDate": "2026-06-01T14:29:03.5130929ZZ",
    "RuntimeId": "db2bb6c7-7781-4e2e-ba76-122d53c3c811",
    "SourceContext": "Pulsar.Client.Api.PulsarClient"
  },
  "Timestamp": "2026-06-01T15:47:01.7764963+00:00",
  "dur_process": 0
}

The error text at position 1 is coming from broker and is the reason why client can't succesfully reconnect and keeps trying. The log is produced from this line (and it is only produced when serverError = ServerError.ServiceNotReady):

pulsar-client-dotnet/src/Pulsar.Client/Internal/ClientCnx.fs

Line 483 in 5a77150

Log.Logger.LogError("{0} Close connection because received internal-server error {1}", prefix, errMsg);