Currently we have PRPC + Robot Manager setup with VM servers as our runtime bots. We schedule our bots to execute from start time -> end time via the robot manager scheduler. The bot will pick up any cases in the queue during that time and process them.
However, as we're exploring the case-management aspect of PRPC, we're wondering if it would be possible to setup a case flow to:
(1) Check a specific workgroup (our default / parking workgroup), fetch the first operator (an idle bot), update it's workgroup to the workgroup associated with this case type
(2) Issue a bot start command
(3) After RobotActivity completes, trigger a bot stop command
This would allow cases to trigger bots on-demand, and not just handle cases during the bot's scheduled window.
We were wondering if this would even be possible with the current PRPC/Bot capabilities, and if so, has anyone done this? Any guidance would be appreciated. We are not experts in PRPC and are still exploring capabilities.
This would be a great benefit for use-cases that are time-sensitive, but do not execute regularly. It doesn't make sense for these to schedule an entire bot for long periods to execute this use-case when it spends most of that time idling.
I think the default handling for this type of case would just be to have a bot idling. The startup time and everything required to get a bot ready would then be added to the processing time if you had to start the bot for every case it processes. Depending on how you structure your work queues, you might not have to have the bot completely idle. It could be processing other work queues within that work group with this one being a higher priority. When work comes in to this one, it would process it as soon as the current work item was completed.
Currently, most of our use-case executions take 1-6 hours to complete, making the 15-30 min stop/start time negligible. We have 2 bot VMs scheduled to execute on various days, for most of that day. Given a usecase that executes 16 hours on Monday, if the user doesn't have everything in place by that Monday, they must wait a week for the bot's next scheduled execution for that workgroup.
Multiple workqueues within a single workgroup can still only support a single deployment package though, correct? So, to have multiple use-cases executing in the same workgroup, we'd have to bundle the use-cases into a single deployment?
So in conclusion, we have 2 options if we want each use-case to function in a more timely manner:
(1) We must have a license for each use-case and leave the bot idling within that use-case workgroup so it will pick up cases on-demand.
(2) Group the use-cases into a single deployment and move workqueues for all grouped use-cases into a single workgroup,
Posted: 10 months ago
Posted: 16 Jul 2020 12:33 EDT
Thomas Sasnett (ThomasSasnett)
Lead Robotics Architect u2013 Pega Consulting
I think that putting them into a single deployment package is the best solution at this time.You might discuss this with your Client Success Manager to see if this can be added as a new feature. It seems like the ability to have a specific case type scheduled immediately based on priority would be a useful feature.