Perhaps you could implement a watchdog style setup. If you use a delay node, you could trigger the delay node every time you sent a message and reset it every time a successful output happened on your MODBUS channel. If you send something into the MODBUS channel and nothing comes out, the delay node can resend the message after a period of time and try again. Looping this back in on itself so the message is continuously retried until successful will ensure the message is eventually sent.
I think the most you would need is adding a delay node in parallel between your messaging node and your MODBUS nodes. On the output of your trigger sender, just add another link that goes to the delay node with a {reset:true} message. This will clear the delay node any time a successful message is received and reset the watch dog. Set your delay for something useful like 5s or some arbitrary amount you know normal messages will clear in but abnormal messages won't.
Make sense?
