Metrics

Overview

vArmor now includes a comprehensive metrics system. This document describes the available metrics, their configurations, and how to enable them in your environment.

Setup Workflow

1. Install prometheus-stack:

   helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
   helm repo update
   helm install prometheus prometheus-community/kube-prometheus-stack

2. Install vArmor with metrics enabled:

   helm install varmor varmor/varmor \
     --set metrics.enable=true \
     --set metrics.serviceMonitorEnabled=true

3. Import Grafana dashboard:

   • Access your Grafana instance
   • Navigate to Dashboards > Import
   • Upload the dashboard JSON file from grafana/panel.json
   • Select the appropriate Prometheus data source
   • Click Import to finish

Enabling Metrics

To enable the metrics system:

1. Set metrics.enable to true in your values configuration
2. To create a Prometheus ServiceMonitor, set metrics.serviceMonitorEnabled to true

Once enabled, metrics are exposed at the /metric endpoint on port 8081 of the manager.

Available Metrics

Profile Processing Metrics

These metrics track the status and performance of the ArmorProfile object processed by the Agent. All profile processing metrics include the following labels:

• namespace: The namespace of the profile
• profile_name: Name of the profile
• node_name: Name of the node

Metric Name	Type	Description
varmor_profile_processing_success	Counter	Number of successful profile processing operations
varmor_profile_processing_failure	Counter	Number of failed profile processing operations
varmor_profile_change_count	Counter	Number of profile changes

Webhook Metrics

These metrics provide insights into admission webhook operations of the Manager.

Basic Webhook Metrics (No Labels)

Metric Name	Type	Description
varmor_admission_requests_total	Counter	Total number of admission requests
varmor_mutated_requests	Counter	Number of requests that were mutated
varmor_non_mutated_requests	Counter	Number of requests that were not mutated

Webhook Latency Metric

The varmor_webhook_latency metric is a histogram that measures webhook processing latency with buckets at 0.1, 0.5, 1, 2, and 5 seconds. This metric includes the following labels:

• request_uid: Request UID
• request_kind: The type of workload be submitted
• request_namespace: The namespace of workload be submitted
• request_name: The name of workload be submitted
• request_operation: Whether the workload be mutated by Manager or not

Grafana Dashboard

A pre-configured Grafana dashboard is available in the codebase for visualizing these metrics. The dashboard provides comprehensive views of both profile processing and webhook performance metrics.

Use Cases

Monitoring Profile Processing

• Track success and failure rates of profile processing across different namespaces
• Monitor profile changes over time
• Identify nodes with profile processing issues

Webhook Performance Analysis

• Monitor admission request volumes
• Track mutation rates
• Analyze webhook latency patterns
• Identify potential bottlenecks in request processing

Technical Implementation Details

The metrics system uses float64 counters and gauges for precise measurements. The implementation includes:

• Float64Counter for cumulative metrics
• Float64Gauge for current state metrics
• Histogram for latency distribution analysis

Integration with Monitoring Stack

The metrics system is designed to integrate seamlessly with the Prometheus ecosystem:

1. Enable metrics in your configuration
2. Configure ServiceMonitor if using Prometheus Operator
3. Import the provided Grafana dashboard for visualization