Axiom Module: Retry Policy Kit (.NET + Polly)

When a dependency gets slow or starts failing, retries can turn a small incident into a full outage, thread pool pressure, socket exhaustion, synchronized retry storms, and cascading timeouts.

Retry Policy Kit is a practical, copy-ready toolkit for shipping safe, bounded resilience in .NET (Polly + HttpClient patterns), with the missing pieces teams usually discover the hard way: stop rules, budgets, telemetry, and rollout guidance.

What you get

• Production-ready retry rules (backoff + jitter + caps + stop rules)
• Timeouts-first guidance and “total time budget” templates
• Circuit breaker and bulkhead patterns to prevent amplification
• A policy registry approach so services converge on one standard
• Observability notes so on-call can answer “what waited?” quickly
• Operator-facing runbooks and checklists for incident response

Included files (Markdown + templates)

• 00-START-HERE.md
• 01-stop-retry-escalate-decision-tables.md
• 02-backoff-and-jitter-defaults.md
• 03-retry-budgets-and-caps.md
• 04-polly-templates.cs
• 05-httpclient-429-retry-after-template.cs
• 06-test-harness-checklist.md
• 07-retry-policy-registry-template.yaml
• 08-retry-event-logging-schema.json
• 09-query-pack.md
• 10-layered-retry-audit.md
• 11-cheat-sheet-one-page.md
• 12-circuit-breakers-and-bulkheads.md
• 13-implementation-notes-registry-and-telemetry.md
• PDF versions included (print/share with your team)

• pdf/00-START-HERE.pdf
• pdf/01-stop-retry-escalate-decision-tables.pdf
• pdf/02-backoff-and-jitter-defaults.pdf
• pdf/03-retry-budgets-and-caps.pdf
• pdf/06-test-harness-checklist.pdf
• pdf/09-query-pack.pdf
• pdf/10-layered-retry-audit.pdf
• pdf/11-cheat-sheet-one-page.pdf
• pdf/12-circuit-breakers-and-bulkheads.pdf
• pdf/13-implementation-notes-registry-and-telemetry.pdf

Formats

• Markdown (.md), C# (.cs), YAML (.yaml), JSON (.json), PDFs (.pdf)

Best for

• .NET / ASP.NET services calling flaky/slow dependencies
• Teams seeing 429s, timeouts, or periodic latency spikes
• Systems where multiple layers retry (SDK + HttpClient + queue + job runner)

Outcomes

• Fewer retry storms and “mystery latency” incidents
• Predictable failure modes (fast fail + escalate)
• Consistent policies across services (less bespoke guessing)

Requirements

• .NET (any recent version) and Polly (examples are Polly-oriented; concepts apply broadly)

License / usage

• License: Single-company internal use. You may use and adapt the templates in your organization’s production systems. Redistribution or resale (in whole or in part) is not permitted.

FAQ (optional)

• “Does this work for legacy .NET?” Yes; templates and patterns are designed to be incrementally adopted.
• “Does it include observability?” Yes; guidance for logging/metrics/spans and what to tag.

If retries have ever made an outage worse for you, this kit gives you a standard you can roll out confidently.