Handle errors and reconnect in C#

The code

SessionReconnectHandler reconnectHandler = null;

session.KeepAlive += (s, e) =>
{
    if (!ServiceResult.IsBad(e.Status)) return;

    Console.WriteLine($"keep-alive failed: {e.Status}");
    if (reconnectHandler != null) return;      // already reconnecting

    reconnectHandler = new SessionReconnectHandler(reconnectAbort: true);
    reconnectHandler.BeginReconnect(session, 5000, (sender, args) =>
    {
        if (!ReferenceEquals(sender, reconnectHandler)) return;
        session = (Session)reconnectHandler.Session;
        reconnectHandler.Dispose();
        reconnectHandler = null;
        Console.WriteLine("reconnected; subscriptions transferred where the server allowed it");
    });
};

// classify before retrying - not every failure deserves one
static bool IsWorthRetrying(StatusCode code) => (uint)code switch
{
    _ when code == StatusCodes.BadCertificateUntrusted   => false,  // needs a human
    _ when code == StatusCodes.BadUserAccessDenied       => false,  // needs a human
    _ when code == StatusCodes.BadServiceUnsupported     => false,  // server cannot do it
    _ when code == StatusCodes.BadSessionIdInvalid       => true,   // transient
    _ when code == StatusCodes.BadConnectionClosed       => true,   // transient
    _ when code == StatusCodes.BadTcpServerTooBusy       => true,   // transient, back off hard
    _ => true,
};

What it prints

keep-alive failed: BadConnectionClosed
reconnected; subscriptions transferred where the server allowed it

Notes

• SessionReconnectHandler is the supported path. It tries TransferSubscriptions first, which preserves your monitored items and their queues.
• Hook the KeepAlive event. It is how you learn the session died promptly instead of at the next read.
• Never retry a certificate or identity error. It will fail identically forever until someone edits a trust list, and a retry loop just buries the log line that would have told you.
• Stagger and jitter reconnects across a fleet. Two hundred clients reconnecting together will finish what the network blip started.
• After reconnecting, confirm your monitored items are still reporting. A transferred subscription is the good case; a silently dead one looks identical from the outside.

What breaks first

Last verified 7 September 2026 against OPC UA 1.05.07. OPC UA Connect is an educational resource and is not affiliated with the OPC Foundation.