UPDATE: Microsoft has rolled out a fix and once again you should be able to apply row filters to your Dataverse flow triggers. I will leave this post here for posterity.
The other day I deployed an update to a Power Automate flow from my development system to my production on a small project. This particular flow ran perfectly fine for many weeks, but suddenly stopped working. On further inspection, I was getting the following issue;
There’s a problem with the flow’s trigger. Fix the trigger.
As I mentioned, the flow had been working fine. I clicked on the Fix the trigger link and found some more interesting information.
Exception parsing avngr_sendcommunications eq 730280001 submitted for attribute filterexpression of callback registration. Target entity: avngr_nicktest. Exception: Microsoft.Crm.CrmInvalidOperationException: Unsupported AttributeType for AttributeTypeCode:FileType
The key interesting aspect was the “Unsupported AttributeType” for a FileType field. No where in my flow did I use the filetype field as a condition or filter.
The table itself does have a FileType field (I really like using FileType fields instead of note attachments or links to Azure Blob storage, another post on that for a later day.) However, it had nothing to do with my flow. The flow is triggered using a choice field;
Thankfully, there is a workaround and it was the suggestion of my wife (who also works with the Power Platform) to use the “old” style of row filters to trigger my flow. I remove the Row filter from the main screen and then choose the ellipses and settings;
And then on the Settings screen add the trigger condition using the syntax we used back in the days before row filters were added to main page of Dataverse (Common Data Service) triggers (old days being a year ago?);
This seems to resolve the trigger issue and once again, you will have a working flow.
Note this is being tracked as a bug and this blog post may soon be out of date if there is a fix being rolled out. From discussions from some folks at Microsoft, the issue is that during the flow trigger registration process, there is some buggy code that doesn’t recognize FileType fields. This is why actions like importing an update make cause the flow to fail while existing flows continue to run fine.
This is normal software development and I hope this blog may end up saving someone some time and hassle.