What would happen a sceduled process catches up with preceding process?

[TetsujiJunicho]

Hi When a process is sceduled as minutely/hourly and a process catches up with the preceding process, what would happen? The one catching will be aborted and then the next one will be started at next round? Regards,

[Denis__Dennehy]

If your process contains objects with a foreground or exclusive run mode then the next scheduled task will fail because the resource is already in use. I am not sure why you would ever schedule a process to run every minute or hour when it is likely to take longer than that to run, it might be worth discussion with your mentor your design.

[Carol.Ouellet]

I'm not op but I am curious about your answer Denis. What if in my case I have a very variable amount of items that can get in the queue at any given time of the day? Would there be a better way to set this up than having the schedule starts every 15min?

[Anonymous]

Well..I have automated a process where the items come in at any given time of the day between 8am til 5pm. The work item flows into an intray. The intray query displays the number of item at a point in time. As the bot processes all the items in the intray, it runs in a loop and queries for more work. This continues throughout the day. So, figure out how does the work item flow into the system and develop your solution around that.

[Anonymous]

Also, when the scheduled process catches up with the preceding process...it should automatically wait until the preceding is complete and then start. I have seen this happening on the AA platform.

[TetsujiJunicho]

Thanks guys. My case might be similar to Carol's and puneetrai's. I hava a scheduled loading of work queue items and a sceduled process consuming these work queue items until they are all marked in the morning, say, at 9am. Besides, during work hours such as between 10am and 5pm, there would be some random loading of work queue items. In order to handle these random work queue items, I think a process should be scheduled cyclically, for example, per 30 minutes/1 hour. Is there any better way to handle this case? Regards,

[Denis__Dennehy]

If you have a very variable number of work items and a very short SLA (i.e. could be 15 minutes work or 1 hours work and an SLA of 1 hour) - then you would need to loot to design a more complex work load balancing solution to allocate robots to your solution when required. That could be some kind of parent process that monitors queues and works the process that is appropriate based upon SLAs/Priorities. Definately one to discuss in more depth with your mentor. In the future roadmap Active Queues will be able to do this kind of work load balancing, but that functionality is not in there yet.

[TomaszAdamski]

@Denis, can't you just add required functionality to a scheduler? It could be postponed for x time (parameter in the schedule) if no license is available. If you have few processes it is not a big deal to coordinate them, but when you have 20-50 of them it becomes a challenge. Regards, Tomek A.

[TetsujiJunicho]

Hi According to ""Work Queues Guide"": https://portal.blueprism.com/system/files/Work%20Queues%20Guide_0.pdf And the section: ""7.4 Real-time Work Requests"" On Page 28: Separate the work request confirmation from the work. With Blue Prism, work will never be completed instantly, it will be queued and processed via the User Interface. The source of the work request should be designed with this in mind. For example, if a work request comes via a Web Service call, Blue Prism should simply add the request to a Work Queue and respond (output) that the request has been received. In the last part, it goes that Blue Prism should ""respond (output) that the request has been received"". How exactly can I respond that the request has been received and trigger a (scheduled) process? Regards,