Documentation

ARCH-Insight Documentation

Learn how to manage Fact Sheets, monitor system health, use chatbot mentions, and interpret diagrams.

System Monitor Admin

The System Monitor provides deep diagnostic visibility into the containers, databases, and message queues comprising ARCH-Insight.

💡 Monitoring Tip The System Monitor runs on a periodic background scheduler. Custom alerts and HTTP diagnostics evaluate automatically on every cycle.

Monitored Services

Database & Connectivity Checks

Monitor Configurations

Monitor Configuration Example

The System Monitor configuration can be imported/exported in the settings tab. Here is an example schema for all infrastructure settings:

monitor_config.json

{
  "ENABLED": "true",
  "LOKI_URL": "http://172.25.30.9",
  "LOKI_TENANT_ID": "",
  "LOKI_QUERY": "",
  "MIMIR_URL": "http://172.25.30.10",
  "MIMIR_TENANT_ID": "",
  "TENANT_IDS": "team-core,team-payment",
  "HEALTH_CHECKS": "[{\"name\":\"Core API\",\"url\":\"http://10.0.0.6:8080/health\"}]",
  "NETWORK_CHECK_HOSTS": "google.com,8.8.8.8",
  "POLL_INTERVAL": "5",
  "HEALTH_CHECK_INTERVAL": "1",
  "NETWORK_CHECK_INTERVAL": "1",
  "THRESHOLD_CPU_HIGH": "80",
  "THRESHOLD_CPU_CRITICAL": "90",
  "THRESHOLD_MEM_HIGH": "85",
  "THRESHOLD_MEM_CRITICAL": "95",
  "THRESHOLD_DISK_HIGH": "80",
  "THRESHOLD_DISK_CRITICAL": "90",
  "ALERT_RECIPIENT_ROLES": "Admin,EA Manager",
  "ALERT_EXTRA_EMAILS": "[email protected]",
  "CUSTOM_ALERT_RULES": "[]"
}

Settings are a flat map of keys (the same ones the Settings tab edits). Credentials such as LOKI_USER, LOKI_PASS, MIMIR_USER and MIMIR_PASS are write-only and are returned masked as "******" on export. The built-in LLM (OpenAI) is configured server-side, so there are no AI keys here; support tickets are handled by the built-in Help Desk (no external ticketing). There is no direct database monitoring — observe data stores via their metrics, logs and HTTP health checks.

Custom Alert Conditions

Beyond the built-in CPU, memory and disk thresholds, admins can define their own alert rules in the System Monitor → Settings tab. Each condition targets a metric, a comparison operator and a value, and fires an alert at the chosen severity when matched. Conditions are evaluated on every poll alongside the standard thresholds.

Metric — one of cpu_usage_pct, memory_usage_pct, disk_usage_pct, http_5xx_rate, pod_restart_rate, network_errors.
Operator — >, >=, <, <= or ==.
Severity — low, medium, high or critical.
Label — an optional human-readable name shown on the alert.

custom_rules.json

"CUSTOM_ALERT_RULES": [
  { "metric": "http_5xx_rate", "operator": ">", "value": 5, "severity": "high", "label": "API errors spiking" },
  { "metric": "pod_restart_rate", "operator": ">=", "value": 3, "severity": "critical", "label": "Crash loop" }
]

Notification Preferences

Each user controls how they are notified from the Notifications screen: toggles for email vs. in-app delivery, and per-type switches for system monitor alerts, survey reminders, task updates and the weekly digest. Preferences are stored per user and respected by the alerting pipeline.

Support Helpdesk Widget

The embeddable contact-form widget that turns visitor messages into helpdesk tickets now lives in the Help Desk section (not System Monitor). Copy the script tag there and paste it before the closing </body> tag of your public site or client portal.