Notification API webhook events - duplicate delivery on architect flow completion

NICE CXone API & Integrations
1

platformClient.api.notifications.api_v2.notification.post is firing events twice for architect flow completions - specifically the flow.completed event. We’ve confirmed the architect flow itself isn’t looping, and the eventBridgeRule.ruleId is the same on both deliveries.

It’s happening consistently - about 70% duplication rate. The eventBody.metadata.eventVersion is also identical on both events; both are v2. Debugging is a nightmare since the Lambda is processing the same completion twice.

notification_id values are different, so it’s not a retransmission of the same event. Something in the notification processing pipeline is clearly broken. Anyone else seeing this?

2

The duplication suggests a misconfiguration in your eventBridgeRule.actions. Specifically, check the action’s runOnCompletion flag - it’s easy to accidentally configure it to fire on both success and failure. You’ll get identical eventBody payloads with that setup. We’re on Zoom Contact Center - haven’t seen this with the platformClient directly, but it’s a common issue when chaining actions.

3 
resource "genesys_cloud_event_bridge_rule" "flow_completion" {
 name = "Flow Completion - Dedupe"
 description = "Dedupe flow completion events"
 event_bridge_rule_id = "your_rule_id_here"

 matching_criteria {
 type = "event"
 event_type = "flow.completed"
 }

 actions {
 type = "web_hook"
 url = "your_webhook_url"
 http_method = "POST"
 run_on_completion = false # important
 }
}

Hey all, is spot on - it’s almost always runOnCompletion. The docs say it fires on both success and failure, but what they don’t tell you is the payload is identical. Which makes debugging…fun.

We’ve run into this a bunch, especially after copy-pasting rules. It’s a quick win to just set run_on_completion = false. You’ll need to handle the success/failure logic in your webhook if that matters, but at least you won’t get hammered with duplicates.

I’m on PT, and honestly, the regional latency on these API calls is killing me - took 10 minutes to import that rule.

Also, double-check the filtering criteria. If you’re doing something weird with metadata, you might be unintentionally matching the same event twice, but I’m betting on the runOnCompletion flag. Ship it.