Fanout service summary

The Pismo platform's Fanout service recovers from both transient and persistent failures without manual intervention. The following information describes what we expect from your endpoint once an event is delivered.

• HTTP/Webhook: The HTTP/Webhook returns a 2xx status code (200-299) to confirm receipt. Your endpoint must respond within 10 seconds. A non-2xx response or timeout causes the event to be re-delivered (with a maximum of up to 10 retry attempts over ~30 minutes).
• AWS SNS: The target subscription is active and the subscriber can process messages at the expected thoughput.
• AWS EventBridge: The target rule and its downstream targets are healthy.
• GCP Pub/Sub: Maintain an active subscription on the target topic with appropriate acknowledgment deadlines.
• Azure Service Bus: The queue or topic has sufficient capacity and is accessible from the Pismo platform.
• Confluent Kafka: Maintain an active consumer group reading from the target topic.

All connectors use the same Fanout recovery mechanisms, including automatic retries, audit logging, credential refresh, and at-least-once delivery guarantees. Connector-specific differences primarily affect how events are transmitted and how delivery failures are identified.