Logic Apps Office Connector becomes unauthorized

Question

I have 100's of Logic Apps in several regions. One thing I do in all my Logic Apps is that I have built in to each error handling to send an alert if any action fails. The alerts are sent using an Office365 connector. Each region will have a connector for alerts, which all the Logic Apps in that region will use. Lately the connector has become unauthorized for no reason. This has occurred three times in the last 5 months. When this happens, I no longer get any alerts and sometimes these are business critical. I've noticed when this happens it will happen to some connectors in one region, not all of them. I have a few others for other reasons othan alerts taht use a different account too. Then a few days later I see it happen again in a different region. They seem to loose authentication in "spurts" slowly across regions. This will occur a few times and not all regions are affected. To reauthenticate is easy, I just have to open the API Connector, click Authenticate and then choose/enter the credentials again. But the lack of visibility is a real problem.

So my questions are:
Is there a way to monitor this? Right now I have to open each connector manually to see. If it's not authenticated anymore, there is banner at the top telling me so.
Is there a way to prevent them from becoming unauthenticated? Until recently I had some running for years without an issue. I guess I need to under stand why to solve this. It looks like Azure is just losing the token or maybe it expires?

Answer 1

@Thomas Johnson

Is there a way to monitor this?

You could setup monitoring alerts for failed actions in general to be notified, which could work for the email system you have now as well except that these emails would lack any custom context.

Is there a way to prevent them from becoming unauthenticated?

Prevention would depend on why they got unauthorized in the first place. It is possible the refresh tokens were revoked by your organization or a password reset could have triggered the same.

You could check the status value in the Resource JSON for more information on why it happened.