Use and Configure Observability Views

The UXP Web UI includes built-in observability views that help you quickly triage issues in your control plane. The dashboard combines resource state with time-series metrics in one place.

note

Time-series metrics require a Standard license. Community edition users can access resource and package state views.

Observability views help with quick fault attribution, not as a replacement for external monitoring stacks.

Observability views can help:

• Identify failing resources. Quickly assess resource health and filter unready resources.
• Troubleshoot control plane issues. Check reconcile rates and API call volumes to validate the scope of an issue.
• Track function performance. Check function execution latency and find slow or failing functions.

For long-term retention, alerting, and deep analysis, use an external monitoring stack (Grafana, Datadog, etc).

Available data

The observability dashboard displays data from two sources:

1. Resource and Package State (QueryAPI)

   • Resource health chart provides a count of resources by Ready condition
   • Package health chart provides a count of providers, functions, and configurations by health status
   • Not Ready resources table filters and inspects resources that aren't healthy
   • Resource creation timeline allow you to see when resources were created over time

2. Time-Series Metrics (Prometheus, Standard editions only)

   • Reconciliation metrics provides reconciliation rates per controller to identify stuck or overloaded controllers
   • External API calls captures the volume of API calls per resource kind to spot providers under heavy load
   • Function latency provides function execution times to identify performance issues

Resources and packages at a glance

The Web UI dashboard provides a quick look at your control plane resources:

Package health

You can click into these views to see more information:

Additional metrics

Feature availability

Observability features vary by edition:

Feature	Community	Standard
Resource health chart	✓	✓
Package health chart	✓	✓
Not Ready resources table	✓	✓
Resource creation timeline	✗	✓
Reconciliation metrics charts	✗	✓
External API call charts	✗	✓
Function latency charts	✗	✓

Prometheus

UXP deploys a lightweight Prometheus instance for Standard license deployments to power the metrics charts. It's preconfigured with 12-hour retention and metric filtering that keeps only what the dashboard needs.

Bring your own Prometheus

If you already run your own Prometheus, disable the built-in instance and point the dashboard at yours through Helm values:

webui:
  config:
    metricsApiEndpoint: 'http://your-prometheus:9090/api/v1'

Your Prometheus needs to scrape UXP components and have these metrics available:

• controller_runtime_reconcile_total
• upjet_resource_external_api_calls_total
• function_run_function_seconds_bucket, _sum, _count