Is there a way to generate an alert when a process has zero tasks for a set duration of time?

Not sure if I understood the question correct. if there are 0 tasks assigned to a user you need a mail or an alert to the administrators. If that is the use case, we can do it with OOTB features available in Appian. We can create a Task Report based on users and check if the count of the tasks are 0 then we can send an email to admins using send email process. Hope this helps :)

Thank you for your response harshav.

Let me try to explain further:

Say we've got a Main Process with subprocesses X,Y,Z -- within X,Y,Z there are numerous tasks that should be executing. Say there's a total of 86 tasks that can be completed over the lifecycle of this process.

If any of the processes stall, due to say a database issue... when we check that process it will have 0 Current Tasks (indicating it's stuck for whatever reason).

Is there a way to monitor that scenario to send an alert?

Process errors alerts can be emailed to a set of users and configure in bunch of combinations. Please explore the process model details alerts tab:

docs.appian.com/.../Alerts_Tab.html

 

The above shows a way to get alerted if process gets stalled because if an issue. There are other ways to get notified as well, let's say if the user is not acting on a task assigned etc can be handled via escalation. Documentation on escalation: docs.appian.com/.../Escalating_a_Task.html

We aren't needing to alert a user because they are sitting on a task too long.

The situation is that the process model errors out on a a database script - for whatever reason, which stalls out the entire process. So at this point, the process isn't moving along. Eventually someone will take a look and see that there are 0 tasks for a process which is the indication that something went wrong. We need to create an alert for that scenario.

Kat,

The alert @shyamb listed is for the process instance, so if your process were to fail (ex. a database call), then the users (most likely application administrators) you configure for process alerts will receive a notification that the process has an error. This alert sounds like it would satisfy your use case.

Sorry for the confusion.

Admin intervention is needed to fix process errors.
And process models can be configured to send out an alert (email) whenever there's any process error to a set of admin groups or admin users automatically.

Sorry for the confusion.

Admin intervention is needed to fix process errors.
And process models can be configured to send out an alert (email) whenever there's any process error to a set of admin groups or admin users automatically.

My mistake in using the term "error out". We don't necessarily get an error...those we can handle and do get alerts for.

What if the process didn't fail, it just stalled out for whatever reason (sometimes for no reason....just happens and you need to restart the task). In that case, there is no error to trigger an alert, it's just sitting there green acting like it's doing something, when in actuality, it's not doing anything.

In our use case, there are numerous times where processes sit a while waiting for data, because sometimes it takes a while to get all the data needed to move forward. Some of our processes can take six months to complete (it's the nature of what we do). However, while it's waiting, there's an active task waiting to be completed. Which is why timers aren't the answer, because each product development cycle is different. Our only indication that something has gone astray is that our process has 0 active tasks after X amount of time.

What we know: we know how many tasks are associated with a given process, so we can cross check against that [ (if activeTasks == 0 && completedTasks < totalTasks) SEND ALERT!] ....are you aware of any function that can check the active task count?

You can leverage the following task functions on a process report. Your limitation would be querying this process report at an interval. An interval of 1 hour will have better performance vs an interval of 1 minute given a high number of processes.

 

Create a process report for all the required process models in context and columns for number of active task and completed tasks.
Create a nightly job and which will query the your process report using query process analytics. identify process with problems and send the list of all such processes in the email to administrator.

> What we know: we know how many tasks are associated with a given process, so we can cross check against that [ (if activeTasks == 0 && completedTasks < totalTasks) SEND ALERT!] ....are you aware of any function that can check the active task count?
- There are ways to do this. As suggested by other practitioners, create a daily/weekly scheduled processes to check and alert the appropriate user/admins etc.

 

My concern is this "What if the process didn't fail, it just stalled out for whatever reason (sometimes for no reason....just happens and you need to restart the task)." There should be no unattended node (any node/task that is not User input task, or has a form for the user to act on) that runs for more than 60 mins and doesn't move the process to error (paused by exception). You have to find the root cause and fix it. 

 

From documentation: "The activity must alost complete within 60 minutes or the process will pause by exception."

docs.appian.com/.../Assignment_Tab.html