sentry-reliability-patterns

# Sentry Reliability Patterns ## Overview Build Sentry integrations that never take your application down via three pillars: safe initialization with graceful degradation, a circuit breaker that stops hammering Sentry when unreachable, and an offline event queue that buffers errors during outages. Every pattern prioritizes application uptime over telemetry completeness. ## Prerequisites - `@sentry/node` v8+ (TypeScript) or `sentry-sdk` v2+ (Python) - A valid Sentry DSN from project settings at `sentry.io` - A fallback logging destination decided (console, file, or external logger) - Understanding of your application shutdown lifecycle (signal handlers, container orchestration) ## Instructions ### Step 1 — Safe Initialization with Graceful Degradation Wrap `Sentry.init()` in try/catch so an invalid DSN, network error, or SDK bug never crashes the app. Track initialization state with a boolean flag. Protect `beforeSend` callbacks with their own error boundary. Create `lib/sentry-safe.ts` with `initSentrySafe()` and `captureError()`. See [graceful-degradation.md](references/graceful-degradation.md) for full implementation. Key rules: - Never let `Sentry.init()` crash the process — wrap in try/catch, set `sentryAvailable = false` on failure - Verify client creation with `Sentry.getClient()` — invalid DSNs silently produce no client - Always log errors locally as baseline before attempting Sentry capture - Wrap user-supplied `beforeSend` hooks in nested try/catch — return...