First, try 3 easy-to-implement conditions.
Something that may help you is working on chrome in incognito mode.
The problem with chrome may be that it takes up too much RAM and logs too much information, incl. in cookies. As for the advice when the environment should be cleared (kill all processes which are required to handle cases), I think that the first symptom to clear should be 1 failed attempt to download the file. I assume that the problem will begin then.
In my processes, I also do variable cleanup in each workflow at the beginning and end of the flow. So that the new iteration of the case has the variables cleared. This procedure is mainly used to care for the state of the Robot’s memory.
If the Robot holds too much data, the RAM consumption increases. It is possible that your process starts to crash after few hours because Chrome and UI Robot are using too much resources. If that’s so. I would even recommend not to run the process continuously, but to run it every 2 hours of work for 2 hours. So that the robot would have continuity, but in addition to handling errors and clearing the environment after an error, you would have the ability to restart environment by clean the RAM that Ui Robot consumes by logging in / out of the station every two hours and freeing up resources. You set these properties in the Trigger:
Finally If there is still a problem, properly rebuild the Process Framework to handle the incorrect operation of Chrome.
The default framework from UiPath may not be sufficient. We have an advanced state transition in our company.
States at our disposal: Initialization, prepare the environment for queuing cases, queue cases, prepare the environment for processing cases, take a case, process, no cases.
If an unexpected robot behavior occurs in the process, the environment restarts and all processes are killed. After 3 same errors, the process stops with fail. We also have timeout support and a Robot that checks other processes for “Heartbeat”.