Hi there,
I made a post the other day where I thought that OpenRPA was losing connection to Node-RED however, that no longer seems to be the case. The problem seems to be that certain Node-RED paths become unresponsive/stuck and the bot cannot receive a response from them.
The bot can hit and receive responses from some paths in the flow except for the paths that run an API call.
The paths seem to become unstuck after I deploy the flow again.
Is there a way to auto kill Node-Red paths or auto redeploy the flow?
Are you “calling” Node-RED from OpenRPA using Invoke OpenFlow?
If you are, there are a few cases where you might never get a response, and the robot will “hang” forever (yeah, I know, we need to add a timeout soon). If Node-RED restarts, or if Node-RED never gets to a workflow out node, no response will ever be sent, and the robot will hang. If Node-RED or OpenRPA loses connection to OpenCore between the request and the response, the robot might not get a response (depending on the case). If RabbitMQ gets restarted.
This is why we heavily push people toward using work item queues instead. You can make this almost as fast as using message queues directly, and it’s transactionally safe, so you will never get into a place where you don’t know what happened or lose work.
Are you reporting a bug in Node-RED?
You keep talking about paths in Node-RED but also keep mentioning OpenRPA, so I’m not sure if you are reporting an issue in Node-RED or OpenRPA. If this issue is in Node-RED, the best suggestion I can give is for you to install the Node-RED debugger ( or just use the debug node ). This might help you locate exactly where the flow stops, and based on that, we can see if this is an issue in a Node OpenIAP wrote or an issue in a node inside Node-RED.