Slow query logging or monitoring is configured

Without slow query logging, a query that degrades from 50ms to 5000ms as the database grows is invisible until users file support tickets or error rates spike. By then, the query has been running slowly for hours or days. ISO 25010 time-behaviour requires that performance characteristics are observable and measurable; a database layer with no query timing instrumentation provides no signal for performance regressions introduced by new code, schema changes, or data volume growth.

Configure Prisma event listeners or a pg query wrapper to log queries over 1 second. For production, send slow query events to your existing observability platform.

// lib/prisma.ts — log slow queries
const prisma = new PrismaClient({
  log: [{ level: 'query', emit: 'event' }]
})
prisma.$on('query', (e) => {
  if (e.duration > 1000) {
    console.warn(`[slow-query] ${e.duration}ms: ${e.query}`)
    // Sentry.captureMessage(`Slow query: ${e.duration}ms`, 'warning')
  }
})

// Custom pg Pool timing wrapper
const origQuery = pool.query.bind(pool)
pool.query = async (...args: Parameters<typeof origQuery>) => {
  const t = Date.now()
  const result = await origQuery(...args)
  const ms = Date.now() - t
  if (ms > 1000) console.warn(`[slow-query] ${ms}ms`, args[0])
  return result
}

For Postgres server-level logging (self-hosted): add log_min_duration_statement = 1000 to postgresql.conf.