Administrative cancellation of management ops is implemented by the thread that is handling the cancellation interrupting the thread that is executing the op. This works, but the thread executing the op can't readily distinguish between interruption do to admin action and cancellation for other reasons such as server shutdown. The OperationContext code detects interruption but restores the interrupted state to the thread, which is valid for the shutdown case. For the administrative cancellation case, it has the negative side effect of producing InterruptedExceptions later on; e.g. when the thread attempts to send data to the remote caller.
This is particularly an issue in domain mode where the "caller" is a controlling process (HC). The InterruptedException has the effect of terminating the connection from the callee to the controlling process.
The operation execution code needs to be able to distinguish between administrative cancellation and other causes of interruption.