Alert Badge #

Introduction #

The Alert Badge is the application-wide status indicator for the alert engine. It sits in the top toolbar and surfaces currently active, unacknowledged alerts at a glance — showing the highest-priority message inline, separate counts for each severity, and acting as the entry point to the full alerts side panel. It complements the live server-health indicator that renders alongside it.

Features #

• Inline preview of the highest-priority newest unacknowledged alert
• Severity-coded counts for CRITICAL (red), WARNING (yellow), and NOTICE (blue), capped at 99+
• Color-coded button that follows the highest active priority (danger / warning / default)
• Bell icon with badge cluster for quick visual scanning
• Sorted order: by priority first, then most recent createdAt
• Click to open a right-hand side panel listing all alerts
• Side panel filter: switch between “Unread” and “All” alerts
• Acknowledge alerts from the panel to remove them from the unread counts
• Hidden when no alerts are present — zero visual noise on a healthy system

How alerts are produced #

The badge displays the output of a backend alert engine built around two layers:

• Triggers watch low-level conditions (e.g., a dataset value changing, or moving into or out of a range). Each trigger has a retentionTime (seconds) — the window during which a fired trigger is considered active. Common evaluators include DatasetValueChangedEvaluator and DatasetValueInRangeEvaluator.
• Rules compose one or more triggers (many-to-many) into user-facing alerts. When all the triggers required by a rule are simultaneously active, the rule emits or extends an alert with a timeoutMinutes lifetime (default 60).

Evaluation runs every 30 seconds on a Quartz-scheduled job (@DisallowConcurrentExecution, so two runs never overlap). Each cycle clears expired triggers, re-evaluates all triggers in batch, then evaluates rules and creates or extends alerts. Re-firing an active rule extends the existing alert’s timeoutAt rather than creating duplicates, so the badge counts remain stable while a condition persists. When timeoutAt passes, the alert disappears from the badge automatically — the system tracks current state only, not a historical event log.

retentionTime vs timeoutMinutes #

In short: retentionTime is the memory window that lets triggers combine into a rule, while timeoutMinutes is the lifetime of the alert that the rule produces.

Use Cases #

• Surfacing live deviations from operational thresholds in the application toolbar
• Quick triage by severity: critical issues call attention without burying lower-priority notices
• Acknowledging alerts as they are reviewed, keeping the badge focused on what still needs attention
• Pairing low-level triggers (e.g., temperature out of range + valve open) into a single composite rule alert