Process much more unreliable when scheduled

I’ve created UI Vision functions to scrape a Windows desktop application. It’s not 100% reliable when I run it by hand but it’s reasonable. I’ve started running it as a scheduled job and it’s really bad. I run the process for multiple areas, which have between 100 and 200 records to scrape, the first one scraped 7 records before stopping and the second failed on the first record, it did start, so got part of the data but never finished the first record. I’ve started this same area now by hand and it’s done 40 records so far without failing.

Are there any tricks to making a scheduled job more reliable? Also, are there any tricks to making a desktop scrape more reliable generally, as it’s pretty random in failures, which is annoying!

Hi, can you post a video of the screen scraping project running? Then we can hopefully tell you how to improve it further. Ideally, on the video we can see both, the scraping in action plus the macro running. This way we can tell what command does what.

Thanks for getting back to me so quick. Unfortunately, the data is very secure, not government levels of secure but secure.

I’ve actually been looking into solutions myself and doing something with onerror looks like the way to go. I have questions on onerror, I’ll create a new post and close this one.

You can also video or screenshot just the macro, so we get a better feeling of what it does. Otherwise it is tricky to give good hints. Of course, please blur out passwords etc.