How can you troubleshoot RPA bots in real-time?

There are many ways of troubleshooting - what’s important is keeping your process ready to be troubleshot when needed 1. Have proper logging mechanism 2. Cover the predefined exceptions with proper Status Code or identification code for the business to easily understand 3. Follow standard framework this helps in auto logging so major errors are covered 4. Now use tools which are available in market to TB - Roboyo Coach can help in troubleshooting

If you have a big process with a lot of workflows and some of them are being reused for different parts of the automation I have found very useful to make a small reference in the log event to the location of the log (i.g.: the name of the workflow or the part of the process), that way whenever the automation breaks anyone can easily find the spot where the error is happening by targeting the range of the last log registered in orchestrator and the next log activity in the project.

Logging and tracing is an important step, but if the underlying apps have changed significantly enough (they often do), to break the bot, you might have to go back to the original app to redefine the selectors / match rules. If the apps are changing frequently (and more dynamic web pages and configurable fat clients do, then either don't use them in the automation OR involve the change control team that your downstream bots will break. Some IT teams won't care if they are constantly solving other problems but you may find better results when all teams are working in synergy.

To troubleshoot RPA bots in real time, monitor bot execution logs for errors, exceptions, or unexpected behavior. Utilize debugging tools provided by the RPA platform to step through the bot's actions and identify issues. Implement exception handling to gracefully manage errors during runtime. Collaborate with the RPA development team to analyze and resolve complex problems. Regularly review and update bot scripts for optimal performance, and leverage logging and reporting features for comprehensive diagnostics.

Detailed Logging is crucial as this and only this gives the actual idea of what might have happened during an unattended execution. Using of different logging levels like trace, info, warning, error and fatal becomes really important while doing so. Also one can take screenshot during error and have it saved locally.

Again, break points and debug helps BUT then you have to fix it and this is often an 90% / 10% rule. 10% debugging time and 90% or more, fixing it. Not all fixes are the same and if they are constantly in bot breakage syndrome, perhaps use a more robust match (x-ray) v selectors.

Using breakpoints and debug mode in RPA development is essential for effective troubleshooting. Breakpoints pause the bot's execution at specific points, allowing you to inspect its state, while debug mode lets you run the bot step-by-step or resume until the next breakpoint. These tools help isolate and test parts of the bot's workflow, verifying logic and functionality. In UiPath Studio, the Locals and Output panels provide real-time insights into variable states and log messages, ensuring accurate and efficient bot operation.

While debugging, if using an IDE like UiPath Studio, not only can you see the state of variables and expressions, but write new expressions in the Immediate panel to validate potential fixes. This speeds up the process of fixing any issues.

To recreate the issue and perform slowly in dev like environment breakpoints and debug mode can be very effective. Again, this is something that can help after error has occurred.

Exception handling and retry are options that you can use in the code to make it more robust and less vulnerable to any minor or one time issue like network outage, unwanted popup, content unavailable, end point unresponsive etc

Retry mechanisms work hand-in-hand with exception handling. If an error occurs, the bot can automatically attempt to retry the failed action after a short delay. For example, if a network error prevents the bot from accessing a database, it can wait a few seconds and try again. This is particularly useful for transient issues that might resolve on their own.

Data validation and testing can help RPA process to not fail in the first place, in terms of troubleshooting I don't see them as very effective but yes they are really good when we want to ensure seamless run of our robot at least from input data perspective.

Most RPA platforms offer built-in validation and testing features. You can set up validation rules within your bot’s workflow to check data as it’s processed. For testing, create a test environment where you can run the bot with different sets of data to see how it performs. Automated tests can also be scheduled to run regularly, ensuring ongoing reliability.

For community support, the UiPath Forum,and Blue Prism Community are great places to start. Don’t forget to check out RPA-related blogs, webinars, and courses for continuous learning.

One such external tool can be a live dashboard that auto refreshes in every 5-10 mins to have status from current production environment. While they can hold crucial business data, at the same time they can hold almost real time status of running robots and failures which can help aid an alerting mechanism.

Another thing to consider when designing the process is are any direct access methods available? Is there an API, can I set up a SQL database to pull data from the source on a regular basis? If you can do direct access you avoid lots of issues that revolve around ui manipulation in addition to aiding in troubleshooting since you can verify the results of a call much more easily.