Robot in VM keeps failing to recognize selectors

After running a process with Chrome and with SAP in Virtual windows 10, robot stops to recognize simple selectors and crashes. It happens every few days, but when I restart the server, it runs smoothly again. There is no increased memory consumption, it does not appear that there any cashed up apps. What could be the reason and solution?

Are you sure that the elements on the application don’t change? Example: If your selector is taking window title into consideration, and if that title for some reason changes, your Robot will fail to recognize the selectors and fail.

Similar things can happen with Window-based applications.

Is there an error message in the log? Did you compare the selector that’s in the error message with the physical state of the application before the crash and determine why the robot is failing to recognizing it?


Thank You for your reply. When I log in to application I need to attach window. It fails sometimes. Selectors do not change. After successfull times it fails. When you restart it is ok. If you click manually on window then Robot recognizes and attaches window instantly.

If clicking on the Window works, can you add a click event that first clicks on the window. This might help your robot and may resolve the issues you’re having.


I have added click image, but still it is a mystery why it happens. Workaround is not a solution anyways.

This is a new one. Unless we see it in action, or the error messages, it would be tough to suggest further remedies.

I suspect there is something wrong with Virtual machine itself. This problem only occurs after many successful attempts and after it occurs - a simple restart is a solution. I think it has something to do with SAP GUI application. RAM consumption is not increased. Something else is preventing robot from doing it’s job.

I can relate to this. One of my projects running on my local machine completely overwhelmed the resources and the robot simply couldn’t proceed. There was nothing wrong with my machine that ran even heavier Java client applications and I had never seen such a situation in the past.
I had to enhance my process to restart automatically based on a set number of iterations after which the robot was able to get through the entire process end to end.

If you implement something like this and that resolves your issues, it would prove the point.

In the past we had similar problem when IE suddenly started to crash after some time. Turns out IE was cashing and it was reflected in memory consumption, so we just restarted IE after some set number of iterations. But this time it is with machine. Will let know if figure it out.

Hello @NotFranmax. We ran into some similar issues, did you figure how to resolve the problem ?
Sorry for the topic unearthing.