How to monitor Laravel in production

Five layers cover 95% of what matters: (1) exceptions with fingerprinted grouping and alert channels, (2) slow SQL queries with pattern-level p95 trending, (3) queue workers, attempts, and per-class failure rates, (4) scheduled task invocations with missed-run detection, (5) per-route latency with p95 trending. Logs and outgoing HTTP are secondary layers. Infrastructure-level uptime is table stakes — covered by a separate uptime tool.

Laravel Telescope is free but dev-only. Flare starts at $9/mo Hobby (single user/project, 10K errors) and covers exceptions only — you still need something for queries, jobs, and schedules. The cheapest full-APM option is NightOwl from $5/mo flat (Hobby, 1 app) or $15/mo Team (3 apps) with BYOD Postgres; it doesn't scale with traffic. Self-hosting OpenTelemetry + Grafana is free in dollars but expensive in ops time.

Partially. Laravel ships plenty of hooks — DB::listen, event listeners for queue/notification/mail events, the scheduler's onSuccess/onFailure callbacks. You can write everything to your database and build dashboards with Livewire or Filament. It works, but you're committing to maintain a monitoring product alongside your actual product.

Exceptions. If an exception fires in production and nobody sees it, you have zero visibility. Start with either the free Flare tier or a basic NotifyOnException Slack channel, then add performance and queue monitoring once you have baseline error coverage.

Trigger a synthetic error in staging or behind a feature flag (a controlled exception) and verify it surfaces in your monitoring within a minute or two. Do the same for a slow query (add a SELECT pg_sleep(2)) and a failing queued job. Any tool that can't reliably surface synthetic failures within its expected latency window isn't ready for production.

Both. Logs capture free-text events — good for auditing and ad-hoc debugging. APM captures structured per-request telemetry — good for aggregation and trending. They answer different questions. A tool like NightOwl does both; or you can run an APM plus a log aggregator like Loki or Better Stack.

Monitoring answers pre-defined questions: 'is the /checkout route slow?'. Observability lets you ask arbitrary questions after the fact: 'which users experienced high latency on Tuesday between 2 and 4 PM?'. Classic APMs lean monitoring; OpenTelemetry + high-cardinality trace backends lean observability. For most Laravel teams, monitoring is sufficient.

Minimum 14 days for trend comparisons (week-over-week debugging). 30-90 days covers most post-mortems. SOC 2 and similar require 365 days for audit logs (a subset of all monitoring data). BYOD storage like NightOwl's Postgres model lets you keep as much as your database has room for — unlimited in principle.