I’ve got a Webhook connection(or rather two) sending to two different Power Automate flows, one for Transaction failures and one for Jobs faulting.
The flow sends messages to a Teams channel so that whoever is on the support rotation can see when failures trigger.
However the Transaction Failure hook seems to be sending two alerts every time a transaction fails. The Power Automate flow receives two identical messages, the only differences are the OrganizationalUnitID sub-second timestamp difference and some request identifiers in the headers.
I’m pondering that this could be a result of having the process in one folder and linked in another, but I cannot readily prove that this is the case.
I’m looking for any ideas as to why a process failing a transaction sends two identical messages because this turns our alerts feed into basically 50% noise and degrades its utility considerably.
This is the snap of the subscribed events for both hooks I’m using, there’s only one for each because I have to have separate Power Automate flows to break each one down.
To continue this discussion, based on observing the OrganizationUnitID from the TransactionFailure messages that come in to Power Automate, I was able to filter out one specific ID that seems to correlate with the Shared folder that we link many of our processes to(helps with support).
Filtering out these messages seems to have reduced the duplicates although we’ll need to run this for a while to determine if we’re losing anything else.
My theory is that because the processes are linked in a folder besides the one they’re already in, they are seen twice by whatever is triggering the alerts on webhooks.
This solution will definitely cause missed alerts if a process is ever built directly in the shared folder, plus it’s just not ideal to have the PowerAutomate using hardcoded “filters” on traffic, but at least it’s stopped the ridiculous overabundance of messages.