Refactor/observability

[LeonardoVieira1630]

Summary

Introduces end-to-end observability across all four services (consumers, dispatcher, logic-system, subscription-server) by wiring in the shared @anticapture/observability package. Each app now bootstraps OpenTelemetry tracing, emits structured logs via an injected pino logger instead of console.*, and the subscription-server exposes a Prometheus /metrics endpoint for scraping. Existing behavior is unchanged — this is a cross-cutting instrumentation refactor.

Changes

• Add @anticapture/observability + @opentelemetry/api dependencies to every app and pino-pretty at the root for local log formatting.
• Add a side-effect instrumentation.ts bootstrap in each app that calls createObservabilityProvider(<service>); imported first in every index.ts so tracer patching happens before other modules load.
• Wrap I/O-facing classes (clients, repositories, handlers, services) with wrapWithTracing(...) at composition roots in each app's index.ts.
• Replace console.log / console.error usages with a structured createLogger(<service>) instance injected through constructors across clients, services, repositories, triggers, webhook server, and Fastify error handler.
• Add a /metrics route on subscription-server that serves Prometheus output via the exporter exposed from instrumentation.ts.
• Add try/catch around app startup in each index.ts to log failures through the structured logger before exiting.
• Delete legacy per-app logger files in favor of the shared observability logger.
• Make timestamps deterministic in non-voting-handler.test-factory.ts and update the related trigger-processor test accordingly.
• Add pnpm.onlyBuiltDependencies: ["sqlite3"] to the root package.json.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
flow-editor	Ready	Preview, Comment	Apr 24, 2026 6:42pm
notification-system-dashboard	Ready	Preview, Comment	Apr 24, 2026 6:42pm