Skip to content

Using Grafana

  1. Open metrics.o.webex.com.
  2. Select the WxCDI folder and the relevant device/service dashboard.
  3. Set a time range that includes the symptom and a known-good period before it.
  4. Filter from region → customer/tenant → cluster → device where those variables exist.
  5. Check the latest sample timestamp and compare a healthy peer.

Reading Alerts

State Meaning
Normal The query condition is not currently met; still confirm data freshness
Pending The condition is met but the configured pending period has not elapsed
Alerting The condition and duration are met
No Data The query returned no series; investigate loss of visibility
Error Rule evaluation failed; validate service health independently

Creating Panels

Use Explore first. Confirm the metric and labels across representative regions, then add the panel to an approved dashboard. Define units, legend, time range, no-data behavior, ownership, and source-control location.

Useful PromQL Queries

up{job=~".*<device_or_service>.*"}
min_over_time(up{job=~".*<device_or_service>.*"}[15m]) == 0

Inspect actual label values before adding customer, region, or device filters.

Tips & Tricks

  • Avoid “Last 30 days” for first-response queries; start narrow and expand.
  • A green panel with old data is not healthy evidence.
  • Use dashboard/search links with the time range and filters in incident handoffs.
  • Do not overwrite shared dashboards with temporary investigation panels.