Laravel Logging

Logging is the observability floor — Log::info/Log::error with context — routed through channels to files, syslog, or Slack, so a production incident has a trail.

Log::error('Payment failed', ['order_id' => 7])
   ↓  (the channel stack)
   └─► storage/logs/laravel.log      (daily file)
   └─► syslog                        (system logs)
   └─► Slack                         (alert the team, via webhook)

The checklist's example is the shape: message + context. The message says what; the context says which order, which user, which request. Context is what turns a log line from noise into an investigation.

The levels

Log::debug(...)     // dev detail
Log::info(...)      // normal events — order placed
Log::notice(...)    // notable but fine
Log::warning(...)   // recoverable problem — retry happening
Log::error(...)     // something failed — payment declined
Log::critical(...)  // app-level failure — the DB is down
Log::alert(...)     // immediate action needed
Log::emergency(...) // system unusable

Context — the checklist's example

Log::error('Payment failed', [
    'order_id' => $order->id,
    'user_id' => $order->user_id,
    'gateway' => 'stripe',
]);
// the context is what makes the line actionable:
// "which order? which user? which gateway? what did the provider say?"

Channels & stacks

// config/logging.php
'channels' => [
    'stack' => [                       // a STACK: send to several at once
        'driver' => 'stack',
        'channels' => ['daily', 'slack-alerts'],
    ],
    'daily' => ['driver' => 'daily', 'path' => storage_path('logs/laravel.log'), 'days' => 14],
    'slack' => ['driver' => 'slack', 'url' => env('LOG_SLACK_WEBHOOK_URL'), 'level' => 'critical'],
    'syslog' => ['driver' => 'syslog'],
],

Stacking by severity

The classic stack: everything to the daily file, errors and up also to Slack. One Log::error() lands in both — the file keeps the full trail, Slack gets the alert.

'stack' => [
    'driver' => 'stack',
    'channels' => ['daily', 'slack'],
    // slack channel is configured with 'level' => 'error', so only
    // errors/critical go to Slack; info/debug stay in the file
],

Q1. How do you log in Laravel?

Through the Log facade — Log::info('order placed', ['order_id' => $order->id]). The message says what happened; the context array says which order, user, and gateway. Logs route through the configured channel stack — usually a daily file, with errors also forwarded to Slack or syslog.

Q2. What are the log levels?

debug, info, notice, warning, error, critical, alert, emergency — in increasing severity. A channel configures the minimum level it accepts ('level' => 'error'), so debug noise stays in the file while only real problems reach Slack.

Q3. What is a channel stack?

A channel that forwards to several others — 'channels' => ['daily', 'slack']. One Log::error() hits both: the daily file keeps the full history, Slack gets the alert. Stacks let you fan out by destination and severity in one config.

Q4. What should go in the context?

Everything needed to investigate without re-running: the record id (order_id), the actor (user_id), the service (gateway), and any provider response. A log line without context is a "something failed" with no follow-up — context is what makes it actionable.

Q5. What shouldn't go in logs?

Secrets and PII — passwords, tokens, full credit-card numbers, raw user data. Logs are copied, retained, and sometimes shipped; treat them as a leak surface. Log the id, not the sensitive value (Lesson 37's posture).

Senior follow-up: Logging vs metrics vs tracing — when is a log the right tool?

Logs answer "what happened on this specific request?" — the event trail. Metrics (throughput, p95, error rate) answer "is the system healthy?" — aggregated. Tracing answers "where did time go in this request?" — across services. The senior answer: all three, structured — logs with context, metrics for thresholds, traces for latency — and logging alone can't tell you the system is slow.

❌ Log::error('Payment failed') with no context — an unactionable line.

❌ Logging secrets/PII — a leak surface in a rotated file.

❌ dd()/dump() left in code — that's debug, not logging.

❌ All log levels going everywhere — the stack with severity thresholds is the point.

• Log::error('message', ['context' => ...]) — message + context
• Levels: debug → info → warning → error → critical → emergency
• Channels: daily (files) · stack (fan out) · slack (alerts) · syslog · null (tests)
• Stack + severity: everything to file, errors+ to Slack
• Never log secrets/PII — logs are a leak surface
• Logs = event trail; metrics = health; traces = latency — use all three