Observability is infrastructure, not an add-on

Standardize logs, metrics, and traces per domain so production incidents point to the right team.

NestJS·Runtime Resilience·Medium·Since 2026-05-01

nestjs-resilience-observability-infrastructure

Why it matters

Failure modes if this rule is ignored

Stake	If ignored
Blind in prod	When something breaks, the first question is "which domain?" If the logs don't know — you're blind. Without a standard, you end up with five log formats and can't search anything.

How to fix

Logs, metrics, and traces come out of one shared platform library. Every domain uses it — does not roll its own. Every log carries domain and requestId by default.

Examples

Bad

// in one domain
console.log('site created', siteId);

// in another
logger.info({ event: 'agent_started', agent_id: agentId });

// in a third
console.log(`[STUDIO] running agent ${name} at ${new Date()}`);

Good

// libs/shared/observability/src/lib/logger.ts
export function createLogger(domain: string) {
  return baseLogger.child({ domain });
}

// in a domain
import { createLogger } from '@acme/shared/observability';
const log = createLogger('sites-management');

log.info({ event: 'site.created', siteId });

Contribute

View sourceSuggest editDiscuss