Alerts & Status Pages

Alerts and status pages are independent systems in Sazabi, but they can be linked together for automatic issue management. Understanding how they interact helps you build effective incident workflows.

Independent systems

Creating a status page issue does not automatically create an alert, and creating an alert does not automatically create an issue. They serve different purposes:

• Alerts: Notify your team about problems and trigger delivery channels (Slack, email, webhooks, incident.io)
• Status pages: Display component health and track incident history

You can use either system independently:

• Create issues without alerts (for planned maintenance or minor issues)
• Create alerts without status page updates (for internal debugging)
• Link them together for full incident automation

Linking alerts to components

To connect an alert to a status page component, include the statusComponentId when triggering the alert. This tells Sazabi to:

1. Create or update an issue for that component
2. Change the component's status based on alert severity
3. Auto-resolve the issue when the alert is resolved

How to link

When Chat triggers an alert, it can include a status component reference:

trigger_alert({
  name: "Elevated errors on checkout service",
  severity: "high",
  statusComponentId: "checkout-service-id",
  whatHappened: "Error rate exceeded threshold on checkout endpoints",
  whyItHappened: "Database connection pool exhausted",
  howToFix: "Scale up database connections or reduce traffic"
})

The statusComponentId must reference an existing component in the project. If the component does not exist, the alert creation will fail.

Auto-created issues

When an alert is triggered with a statusComponentId, Sazabi automatically creates an issue for that component. The issue:

• Links to the alert for context
• Has a severity mapped from the alert severity
• Appears on the status page immediately
• Updates the component's displayed status

Severity mapping

Alert severity maps to issue severity as follows:

This mapping reflects that low and medium alerts typically represent partial issues (degraded), while high and critical alerts represent complete failures (outage).

Multiple alerts

If multiple alerts are created for the same component:

• Each alert creates a separate issue
• The component shows the worst status (outage beats degraded)
• All issues must be resolved for the component to return to operational

Auto-resolution

When an alert is resolved, its linked issue is automatically resolved as well. This creates a seamless workflow:

1. Problem detected, alert triggered with statusComponentId
2. Issue created, component shows degraded/outage status
3. Problem fixed, alert resolved
4. Issue auto-resolved, component returns to operational

Partial resolution

If a component has multiple active issues:

• Resolving one alert resolves only its linked issue
• The component status reflects remaining active issues
• All issues must be resolved for full operational status

When to link alerts to components

Link alerts to status components when:

• The alert represents a customer-facing issue
• You want automatic status page updates
• The problem affects a specific, trackable service
• You need historical tracking of incidents per component

Do not link alerts when:

• The issue is purely internal (debugging, development)
• No suitable component exists yet
• You want to notify the team without public status impact

Example workflow

Here is a typical flow with linked alerts and status:

1. Detection: Chat detects elevated errors on payment-gateway
2. Alert: Assistant triggers alert with severity high and statusComponentId for payment-gateway
3. Issue created: Payment-gateway shows "Outage" on status page
4. Investigation: Team investigates using the linked alert thread
5. Resolution: Team fixes the issue, assistant resolves the alert
6. Auto-resolution: Payment-gateway issue resolved, status returns to "Operational"

The status page timeline now shows the outage period, and uptime metrics are updated accordingly.

Next steps