Skip to content

Add a New Device Type

Use this process for a new device role or a material change to the collection method. A single new host normally follows the shorter inventory update workflow.

Prerequisites

  • Confirm service and device ownership.
  • Define supported regions and source inventory.
  • Identify stable device_type, node_type, region, data-center, device, and customer labels.
  • Define required metrics, events, dashboard, service alerts, and scrape/ingestion-down coverage.

Step 1: Add to Inventory

  1. Update the authoritative source inventory.
  2. Add device and region mappings in config/devices.yaml and config/regions.yaml if needed.
  3. Run the matching wxcdi-inventory-files generator.
  4. Regenerate syslog metadata, trap metadata, and the hosts cache.
  5. Review that only the intended targets changed.

Step 2: Configure Monitoring

Use an existing SNMP module/exporter when it satisfies the requirements. For a new exporter, define stable metrics, timeouts, error isolation, self-metrics, tests, secure credential loading, inventory consumption, chart/runtime values, and target discovery.

Verify target discovery, successful collection, Prometheus scrape, Mimir remote write, and Grafana query before claiming coverage.

Step 3: Configure Logging

Define source-side syslog/trap/file configuration, regional ingress, metadata enrichment, Splunk instance, index/sourcetype, data classification, and source-health monitoring. Test with non-sensitive representative data.

Step 4: Create Dashboards

Create service-health, resource, dependency, and telemetry-freshness panels. Use current labels, meaningful no-data behavior, source control, and ownership/runbook links.

Step 5: Configure Alerts

Add actionable service alerts and a scrape/ingestion-down alert. Test firing, grouping/suppression, notification delivery, acknowledgement, and recovery.

Step 6: Update Documentation

Update the matching observability-audit YAML and the existing device page in this portal. Do not mark the device covered while the dashboard, alerts, or loss-of-visibility control remain undefined.