Unhandled ObjectDisposedException during driver use

[CSymes]

Describe the bug
In normal operation, our API container app's graph queries began to fail with an ObjectDisposedException, stack trace:

System.ObjectDisposedException: Cannot access a closed Stream.
Object name: 'NetworkStream'.
   at AsyncTransaction.TransactionConnection.OnErrorAsync(Exception error)
   at Connector.DelegatedConnection.SyncAsync()
   at Protocol.BoltProtocolV3.RollbackTransactionAsync(IConnection connection)
   at Internal.AsyncTransaction.RollbackAsync()
   at Internal.AsyncTransaction.RollbackAsync()
   at <>c__DisplayClass54_0`1.<<RunTransactionAsync>b__1>d.MoveNext()
--- End of stack trace from previous location ---
   at Internal.AsyncRetryLogic.RetryAsync[T](Func`1 runTxAsyncFunc)
   at Logging.DriverLoggerUtil.TryExecuteAsync[T](ILogger logger, Func`1 func, String message)
   at Internal.Driver.ExecuteQueryAsyncInternal[T](Query query, QueryConfig config, CancellationToken cancellationToken, Func`3 cursorProcessor)
   at Internal.Driver.ExecuteQueryAsyncInternal[T](Query query, QueryConfig config, CancellationToken cancellationToken, Func`3 cursorProcessor)
   at Internal.Driver.GetRowsAsync(Query query, QueryConfig config, Action`1 streamProcessor, CancellationToken cancellationToken)
   at Internal.ReducedExecutableQuery`3.ExecuteAsync(CancellationToken token)
   at OurNamespace.OurGraphManager.RunGraphQueryAsync(String query, Dictionary`2 parameters)
...

In our code, we open and persist a driver connection as:

...
_driver = GraphDatabase.Driver(uri, AuthTokens.Basic(configValues.Username, configValues.Password));
await _driver.VerifyConnectivityAsync();

and then make calls using this instance throughout the lifetime of the application:

IDriver conn = await GetConnectionAsync();

var results =
    await conn
    .ExecutableQuery(query)
    .WithParameters(parameters)
    .WithMap(record => GetNodeFromRecord(record))
    .ExecuteAsync();

To Reproduce
Unfortunately I have been unable to reproduce this on demand, as it happened seemingly without obvious cause.

Expected behavior
DB connection continues to work as usual.

The stack trace indicates that the fault is internal to the driver. Even if the server connection had failed, the connection pool should handle this and replace it with a new connection instead of multi-disposing any resources.

Version Info (please complete the following information):

• .NET Version: .NET 8
• .NET Driver Version: v5.28.1
• Server: non-Neo4j, Bolt-compatible server

[AndyHeap-NeoTech]

Hi CSymes. That is an unusual one. Is your application multithreaded and sharing the driver instance across threads? I'm wondering if the driver instance could have been closed whilst this query was running?

[CSymes]

The application is multi-threaded, however per the docs, we do hold the driver instance open for the lifetime of the application, so I wouln't expect this to be the issue (all things working as expected).
I'm not seeing this error today, however, so it may well be race-condition related (well, almost certainly). I'll let you know if I manage to repro it again, or if we've accidentally 'fixed' whatever scenario was causing it.

[CSymes]

Re-encountered this today. Upon closer inspection, seems like the setup was that my VPN connection to the server was disconnected by my workstation, invalidating the driver's connection, but then any attempts to use it result in the ObjectDisposedException rather than resulting in a retry or connection-lost error.