Hi,
we need to centralize the opening of an incident (on an external system) following an error found within the Appian platform.
From the documentation it emerges that it is impossible to recover the information of a process that has failed in an unhattended node.
In the various support functions the tasks with problems are counted, but the unhattended nodes are not shown.
Furthermore, the process status is active. Is there some method that we have not taken into consideration?
Unfortunately Appian Support was unable to resolve this issue.
Discussion posts and replies are publicly visible
You can recover the error it in a programatic way but using Appian API.....
There is an AppMarket plug-in that might help you get some info: Appian Error Function
The challenge will be programmatically responding to the error event and using that to call your external incident management system. One option might be to start a "monitor" process instance for every process you start that has an internal clock that when it times out it queries the real process for the error details and then sends those to your incident system. This isn't ideal as you'll be doubling the number of process instances in your environment.
Another approach might be to write an entry to a database table for each started process, and then when each process ends to removes (or marks as complete) that entry, and then have a single process that regularly queries that database to find process instances that didn't complete, and then raises a request in your incident system. Better than the previous option, but still not ideal as you're inferring from data rather than responding to an actual event.
Finally you might want to use the fact that when process models encounter errors you can send alerts to various recipients. One recipient could be a dummy user whose email address is a mailbox that forwards the email to Appian and triggers and Appian process that then queries for the error and sends the details to your incident system. Better, but clunky.
Hi Stewart,
thanks for your feedback.
My opinion is to not use plug-in, I think that this is a very basic functionality, and Appian has this, otherwise you won't see process errors in process monitoring.
My ideas were quite similar to yours:
- a process that every day check if a process is currenlty running (whitout knowing anything about the reason it is still active, but anyway...),
- your last proposal, about sending an email to an email address that triggers the start of a process. For this solutions, would you have any doubts about the safety of this solution? Our idea would be to check the sender and filter out all unknown senders.
Thanks,
Dalila
Hi David,
could you please elaborate futher?