Handling Failures with Messaging
Download as PPTX, PDF
4 likes1,337 views
This document discusses different types of failures that can occur in distributed systems and messaging architectures. It categorizes failures as either transient or permanent, provides examples of each, and demonstrates approaches for handling failures synchronously and asynchronously. The key approaches include retrying transient failures, logging permanent failures, and using messaging to decouple failure handling from the frontend for better retry capabilities and failure replay.
Handling Failures with Messaging
- 1. Elton Stoneman Microsoft MVP & Pluralsight Author http://particular.net Handling Failures with Messaging
- 2. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Types of failure Demo app Synchronous failure handling Asynchronous failure handling Messaging architecture Outline
- 3. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Distributed architecture Timeout / process overloaded Temporary should retry Transient Failures Database Network timeout Pool exhaustion REST API Connection timeout 503 'Service Unavailable'
- 4. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Contract changed Authorization revoked Permanent should not retry Permanent Failures Database Procedure change Permission change REST API 400 'Bad Request' 401 'Unauthorized'
- 5. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Demo App Web App REST APIs Reliable (200) Unreliable (200 or 503) Broken (400)
- 6. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Synchronous processing API calls with WebClient Exceptions from non-OK result App V1
- 7. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Demo V1 No failure handling Bubble up to user Transient & permanent
- 8. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Handling Failures Type? Audit End Retry OK? Failure Transient Permanent Yes No
- 9. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Handling Failures End OK? Failure Transient No Yes Permanent Transient or permanent? Retry policies Audit process Type? Audit Retry
- 10. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Simple failure handling Determine failure type Retry transient; log permanent App V2 try {} catch {}
- 11. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Demo V2 Always appears successful Retry options limited Audit process basic
- 12. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Business / technical fix Failure backlog Replay process Replaying Failures
- 13. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Asynchronous messaging Web App sends message Handler has retry & audit policy App V3
- 14. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Integration happens offline Wider scope for retry Full details for replay Demo V3
- 15. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Messaging Architecture Message All request data Processing data Queue Ordered store Transactions/ACKs Handler Decouples front-end Can be stopped
- 16. Introduction to ServiceInsight for NServiceBusHandling Failures with Messaging Expect failure Handle failure in the right place Retry & backoff damages front-end Decoupled handlers have more options Persisted messages can be replayed Summary