Retries and Failure Handling

Summary

This page records the verified Workforce Scheduling event boundary and its adjacent Attendance integration behavior.

Audience

Developers, integration engineers, operators, support engineers, QA engineers, and solution architects.

Reference Content

Two retry layers are verified.

Layer	Behavior	Exhaustion
Attendance outbox relay	Fixed-interval polling; failed rows remain eligible; retry count increments	No retry ceiling or producer dead-letter transition found
Integration orchestrator	Bounded exponential backoff per configured downstream target	Persists authoritative failed record and attempts a secondary dead-letter publish

The broker consumers use manual acknowledgement. Unexpected infrastructure exceptions reject and requeue a delivery. The Integration dispatcher returns a handled dead-letter outcome after exhausted downstream retries, allowing the original delivery to be acknowledged.

Error text is persisted in bounded fields for the outbox and failed-event records. Operational replay tooling, scheduling-specific replay, poison-message quarantine for the producer outbox, and retry alert thresholds require confirmation.

Source References

• microservices/src/attendance-service/Infrastructure/OutboxRelayHostedService.cs
• microservices/src/shared-kernel/Messaging/OutboxProcessor.cs
• microservices/src/attendance-service/Infrastructure/Persistence.cs
• microservices/src/integration-service/Orchestration/RetryPolicy.cs
• microservices/src/integration-service/Orchestration/EventDispatcher.cs
• microservices/src/integration-service/Messaging/IntegrationEventConsumer.cs

Related Articles

• Events overview
• Event limitations
• Event source map

See Also

• Technical documentation
• API documentation
• Database documentation

Keywords

• Workforce Scheduling events
• Attendance integration contracts
• Transactional outbox

Revision Information

• Status: Draft
• Last reviewed: 2026-07-20
• Next review: 2026-10-20