Add "on error" option to all NWC actions Merged

Hi all,

Thank you all, for the idea, additional comments, and votes.

As you will hopefully have seen, the first stage of Error handling was released in June. For more information, see the Error handling help topic.

Our teams have moved onto other high priority features, but will return to enhance Error handling in the future. There are a few ways that we'd like to extend the feature, and this capability is one of the top ideas to consider.

I'll be moving this back into Review, while we go through the next stage of exploration and validation.

Thanks again and regards,

Kate Huynh

Yes, this would be super helpful, especially with our library issue. At least then, we could pause workflow and loop back through some actions if needed until it goes through.

We have issues with document management right now if a user has a document open it says locked for share use or something like that. If this existed, we could resolve our issue among other workflow failure issues that we cannot get around or are caused for no reason that could be identified.

The lack of this functionality is a serious concern. It is not acceptable for a workflow to fail because NWC is unable to handle an error being returned by an API.

We're having this same concern in our organization as well. NWC can support a lot of complexity in the types of workflows designed - meaning the answer can't be as simple restarting the workflow from the beginning. In my opinion, we really need to have a way to define what happens when an action fails. Perhaps provide a toggle on each action to indicate whether a failure should terminate the workflow or not, as well as making the return code and action result available as action outputs. As Workflow designers, we could then evaluate the output to determine the best course of action.

It would also be extremely valuable to be able to restart/skip actions from the instance details page, so that workflows that encounter unexpected issues (3rd party service down for maintenance) could still be managed.

This sort of functionality would be invaluable to us. It's very frustrating designing a fairly hardened complex workflow and seeing a form go through 4 or more levels of approval only to fail on some action where it had to call out to a web service or write to an SQL server and there was a hiccup in internet connectivity at that moment. Whole workflow fails and the initiator has to put in a new form. No real fault of the workflow designer, but it makes the person putting in the form frustrated when it's a more important form that needs to get approved in a limited time-frame.

Being able to programmatically specify what should happen if an action fails would be great. Sometimes, yes, you might just want the workflow to fail. Other times maybe it's not an issue at all and the workflow should continue. Or you might want to be able to build out some actions where it simply waits a few seconds/minutes and tries the action again.