Skip to content
advanced Phase 84 · Observability Advanced

Error Tracking

Error tracking with Sentry, error grouping, error alerting, and error budgets

45m
0 problems
Topic Progress 0%

Sentry Integration

Sentry Architecture

Application → Sentry SDK → Sentry Server → Alert
     │                           │
     │                           ▼
     └─────── Error Event ◀── Dashboard

Magento Sentry Setup

// composer.json
require {
    "sentry/sentry-php": "^4.0"
}

// app/bootstrap.php
use Sentry\SentrySdk;
use Sentry\State\Hub;

SentrySdk::init([
    'dsn' => 'https://key@sentry.example.com/project-id',
    'environment' => getenv('APP_ENV'),
    'release' => MAGENTO_VERSION,
    'traces_sample_rate' => 0.1,
    'before_send' => function (\Sentry\Event $event) {
        // Filter sensitive data
        $event->setExtras(array_filter($event->getExtras(), function ($key) {
            return !in_array($key, ['password', 'token', 'secret']);
        }, ARRAY_FILTER_USE_KEY));
        return $event;
    }
]);

// Capture exceptions
try {
    $this->orderService->process($order);
} catch (Exception $e) {
    \Sentry\SentrySdk::captureException($e);
    throw $e;
}

Breadcrumbs

// Add context breadcrumbs
\Sentry\SentrySdk::configureScope(function ($scope) {
    $scope->setTag('order_id', $orderId);
    $scope->setUser([
        'id' => $customerId,
        'email' => $customerEmail
    ]);
});

// Add breadcrumb
\Sentry\addBreadcrumb(new \Sentry\Breadcrumb(
    \Sentry\Breadcrumb::LEVEL_INFO,
    \Sentry\Breadcrumb::TYPE_DEFAULT,
    'payment',
    'Payment initiated',
    ['amount' => $amount]
));

Error Grouping

Grouping Strategy

Same Error Group:
- Same exception class
- Same stack trace
- Same error message pattern

Different Error Groups:
- Different exception class
- Different stack trace
- Different message template

Custom Grouping

// Set fingerprint for custom grouping
\Sentry\SentrySdk::configureScope(function ($scope) use ($e) {
    $scope->setFingerprint([
        $e->getMessage(),
        $e->getFile(),
        $e->getLine()
    ]);
});

// Group by exception type
$scope->setFingerprint([
    '{{ default }}',
    $e->getClass()
]);

Error Deduplication

// Sentry deduplicates by default
// Same error in same version = same issue

// Custom deduplication
$errorHash = md5($e->getMessage() . $e->getFile() . $e->getLine());

if (!$this->cache->load('error_' . $errorHash)) {
    \Sentry\SentrySdk::captureException($e);
    $this->cache->save('1', 'error_' . $errorHash, [], 300);
}

Error Alerting

Alert Rules

Rule Type        | Condition              | Action
─────────────────|────────────────────────|──────────────
New Issue        | First occurrence       | Email + Slack
Regression       | Fixed issue reappears  | Email + Slack
Frequency Spike  | Error rate > 5x       | PagerDuty
Total Volume     | > 100 errors/hour      | Email

Sentry Alert Config

# .sentry/alerts.yml
alerts:
  - name: "High Error Rate"
    conditions:
      - type: error_rate
        value: 5
        period: 1h
    actions:
      - type: email
        targets: ["ops@example.com"]
      - type: slack
        channel: "#alerts"

  - name: "New Critical Error"
    conditions:
      - type: new_issue
      - type: level_equals
        value: "error"
    actions:
      - type: pagerduty
        service: "magento"

Alert Notification

// Add alert context to errors
$event->setTag('severity', 'critical');
$event->setTag('team', 'payments');
$event->setTag('service', 'order-service');

// Set user impact
$event->setUser([
    'id' => $customerId,
    'email' => $customerEmail,
    'ip_address' => $request->getClientIp()
]);

Error Budgets

Error Budget Concept

SLA: 99.9% availability
Error Budget: 0.1% = 43.8 minutes/month

If errors consume budget:
- 50% consumed: Warning
- 80% consumed: Feature freeze
- 100% consumed: Reliability sprint

Error Budget Calculation

// Calculate monthly error budget
$slaTarget = 99.9;
$errorBudget = 100 - $slaTarget; // 0.1%

$monthMinutes = 43200; // 30 days
$budgetMinutes = $monthMinutes * ($errorBudget / 100); // 43.8 minutes

// Track consumption
$actualDowntime = $this->getDowntimeMinutes();
$remainingBudget = $budgetMinutes - $actualDowntime;
$consumptionRate = ($actualDowntime / $budgetMinutes) * 100;

// Alert based on consumption
if ($consumptionRate > 80) {
    $this->alertService->notify('Error budget 80% consumed');
}

Error Budget Policy

Budget Remaining | Action
─────────────────|──────────────────────
> 50%            | Normal feature development
25-50%           | Increase testing, review deployments
10-25%           | Feature freeze, focus on reliability
< 10%            | Hard freeze, reliability sprint only
0%               | All hands on reliability

Quiz

1. What is Sentry used for?

Question 1 options

2. What is an error budget?

Question 2 options

3. When should feature development freeze?

Question 3 options

Flashcards

Question

Sentry purpose?

Click to reveal answer

Answer

Error tracking, grouping, alerting, and monitoring

Question

Error budget concept?

Click to reveal answer

Answer

Allowed downtime percentage before SLA breach

Question

Budget 80% consumed action?

Click to reveal answer

Answer

Feature freeze, focus on reliability

Question

Error deduplication?

Click to reveal answer

Answer

Group same errors by stack trace and message pattern

Revision Notes

Key Takeaways

  • 1. Sentry provides real-time error tracking with grouping and alerting
  • 2. Error grouping uses stack trace and message patterns
  • 3. Alert on new issues, regressions, and frequency spikes
  • 4. Error budgets guide feature vs reliability trade-offs
  • 5. 80% budget consumption triggers feature freeze

Interview Tips

  • • Explain error grouping strategies and deduplication
  • • Discuss error budget policies and their business value
  • • Describe Sentry integration and alert configuration

Cheat Sheet

Error Tracking:
  Sentry: Real-time error monitoring
  Grouping: Stack trace + message pattern
  Deduplication: Same error = same issue

Alerting:
  New issue: Email + Slack
  Regression: Email + Slack
  Frequency spike: PagerDuty

Error Budgets:
  SLA 99.9% = 0.1% budget = 43.8 min/month
  >50%: Normal development
  25-50%: Increase testing
  10-25%: Feature freeze
  <10%: Reliability sprint