SC in running on its own machine, and the error messages were in the local transactional DLQ of the sending endpoint.
But I cannot reproduce this anymore after restarting the machines where the endpoints are installed. Maybe the connection was cache? I’ll reopen this if I notice the same behavior again.
My first thought would be that MSMQ could not send the heartbeat messages the other machine. If that takes too long, the messages in the Outgoing Queue won’t be delivered at the other machine, the one where Service Control is installed on, and MSMQ will eventually move them to the DLQ.
I’m not 100% sure if disabling the DLQ would disable that feature (moving undeliverable messages in outgoing queue to dlq) in MSMQ, though. Would need to verify that.