Running into a weird bottleneck with our custom agent desktop app’s backend. We’re processing Genesys Cloud interaction events via EventBridge to update local state in near real-time. The Lambda function is set to a batch size of 1, so we shouldn’t be hitting concurrency limits on processing logic.
Here’s the symptom: during peak hours, the Lambda concurrency spikes to the account limit (1000), and EventBridge starts dropping events with THROTTLED status. The function itself executes in ~50ms on average, well under the 3s timeout. I’ve checked the CloudWatch metrics, and the IteratorAge is creeping up, but the ConcurrentExecutions graph looks like a sawtooth wave that never dips below 900.
{
"event": {
"source": "genesys.cloud",
"detail-type": "interaction.created",
"detail": { ... }
}
}
I’ve tried increasing the Lambda provisioned concurrency, but that just burns money without fixing the throughput. The EventBridge rule has a retry policy set, but it’s not helping. Is there a specific configuration on the EventBridge target side that limits the rate of invocation regardless of Lambda capacity? Or is this a known issue with how Genesys Cloud publishes to EventBridge?
Tried:
- Increasing batch size to 5 (didn’t help, just made errors worse)
- Checking Lambda cold starts (using provisioned concurrency, so that’s out)
- Reviewing VPC endpoint limits (not using VPC for this Lambda)
Any ideas on what’s capping the throughput here?