Just how unique are the ppids generated by Appian? At what point would Appian st

AFAIK process IDs are be reused as soon as the process is deleted. So why do you ask? Do you think about using the ID for some sort of integration to another system? In cases I need to identify a process I create a PV and put a UUID in it.

Got it. So the caveat being, the ID is unique within the current live process. So effectively there is a pool of IDs available to the system which are perpetually recycled.
The UUID option looks a little excessive but well worth a look since this would clearly render a unique id.
Thanks, J

Process IDs are not reused. Even when a process is deleted.

Hi Owen
It happened with me as well, when the process id in the environment started repeating occasionally.
We were storing instance Id in the database for some business logic on process instance. However as soon as process instance Id started repeating, the table has duplicate ids which was incorrect as we were under the impression that whatsoever Appian will generate unique process instance ids in all cases.
Any thoughts on this ?

Since in your scenario you were manually storing the process id's in the DB there is a big possibility of a problem in your logic. Since the DB is outside Appian and you're inserting them using a smart service if there's an error in your logic, or someone restarted the node or the node wasn't configured properly there can be inserts that you were not expecting. For this particular scenarios is the responsibility of the developers of your team to debug further; in terms of Appian's architecture process IDs are not re-used.

On my system there are 3 exec engines. The process IDs started at about 1 and 268.435.000 and 536.871.000. So what happen when the IDs of the first exec engine reach the number range of the second? I know that this is a very big range and a system would need to start new process on a high frequency for many years. But from my view as a computer scientist this is not something I would rely on.

Thanks for the clarity Owen and Eduardo and for your input Stefan.
In which case I guess you have to be careful when storing to an external system as clearly there is a risk of data replication if incorrectly handled. Also would the type need to be set as a bigint when in SQL? Or would standard int be sufficient when looking at long term storage of process data? When considering your answer we do have a high throughput of transactions in our system which is now four years old running in 8 exec engines. I am sure there are bigger systems out there amongst the customer base so perhaps there is a better use case to draw upon.

I'd say bigint would work better.

So do you mean After the process is completed and archived , the processID would be released back the pool and probably be reused again?

As Eduardo and Owen stated, process ids are never reused.