Performance Insights

Monitor cluster performance with metrics and insights for capacity planning and optimization.

What We Monitor

Performance data with context to help you make informed decisions about your infrastructure:

• Server Health: Infrastructure resource utilization
• Application Performance: Workload efficiency metrics
• Resource Usage: Capacity and optimization opportunities
• Capacity Planning: Growth trends and recommendations

Tip: Metrics with Context

Performance metrics include descriptions and recommendations to help you understand what actions to take.

Server Health (Node Monitoring)

CPU: Is Your Server Keeping Up?

What we show:

• Clear health status: "Healthy", "Under pressure", or "Overloaded"
• Usage patterns: consistent, spiky, or growing trend
• Impact on your applications

Example insights:

• ✅ "CPU usage is healthy at 45%. Your server has plenty of capacity."
• ⚠️ "CPU usage is at 85%. Your server is working hard but managing. Consider scaling if usage continues to grow."
• 🔴 "CPU usage is at 95%. Your applications may be slow. Add more nodes or reduce workload."

Memory: Do You Have Enough RAM?

What we show:

• Available memory in simple terms
• Memory pressure warnings before problems occur
• Which applications are using the most memory

Example insights:

• ✅ "Memory usage is healthy at 60%. Plenty of room for growth."
• ⚠️ "Only 15% memory free. New pods may have trouble starting. Consider adding nodes."
• 🔴 "Memory is critically low at 95%. The server may start killing processes. Immediate action needed."

Disk Space: Running Out of Room?

What we show:

• Free disk space in GB/TB (not just percentages)
• Growth rate: how quickly you're filling up
• Which areas are using the most space

Example insights:

• ✅ "Disk usage is at 45% (220 GB free). Plenty of space available."
• ⚠️ "Disk is 88% full (30 GB free). You'll run out of space in approximately 2 weeks at current usage."
• 🔴 "Disk is 96% full (8 GB free). Critical: Pods may fail to start or write logs."

Network Metrics

Network Throughput

• Inbound traffic (bytes/sec)
• Outbound traffic (bytes/sec)
• Network errors and drops
• Connection statistics

Load Average

System Load

• 1-minute load average
• 5-minute load average
• 15-minute load average

Alert Thresholds:

• Warning: Load5 > CPU count * 0.8
• Critical: Load15 > CPU count * 0.8

Pod Monitoring

Resource Consumption

CPU Usage

• Current CPU consumption
• CPU requests vs actual usage
• CPU limits and throttling

Alert: Pod CPU throttling > 25%

Memory Usage

• Current memory consumption
• Memory requests vs actual usage
• Memory limits
• Out-of-memory (OOM) events

Alert Thresholds:

• Warning: Memory usage > 90% of requests
• Critical: Memory usage > 95% of limits

Pod Health

Status Monitoring

• Pod phase (Running, Pending, Failed, etc.)
• Container readiness
• Liveness probe status
• Restart counts

Alerts:

• Pod not ready > 5 minutes
• Excessive restarts (> 5 in 1 hour)
• Pod OOM killed
• Pod errors

Container Metrics

Per-Container Statistics:

• CPU usage per container
• Memory usage per container
• Restart history
• Exit codes and reasons

Workload Monitoring

Deployment Status

Replica Monitoring:

• Desired replicas
• Current replicas
• Available replicas
• Unavailable replicas

Alert: Deployment replicas unavailable

StatefulSet Status

Ordered Pod Monitoring:

• Desired replicas
• Current replicas
• Ready replicas
• Pod management status

Alert: StatefulSet replicas unavailable

DaemonSet Status

Node Coverage:

• Desired pods (one per node)
• Current pods
• Available pods
• Unavailable pods

Alert: DaemonSet replicas unavailable

Volume Monitoring

Persistent Volume Claims

Storage Metrics:

• Volume capacity
• Volume usage percentage
• Available space

Alert Thresholds:

• Warning: Volume usage > 90%
• Critical: Volume usage > 95%

Alert: Volume stats missing (unable to retrieve metrics)

Ingress Monitoring

Traffic Metrics

Request Statistics:

• Request count
• Request rate (requests/sec)
• Response latency (p50, p95, p99)

Alert: High request count (unusual traffic spike)

Error Rates

HTTP Status Codes:

• 2xx success rate
• 4xx client error rate
• 5xx server error rate

Alert: 5xx error rate > 5%

Request Latency

Response Time Monitoring:

• Average request latency
• 95th percentile latency
• 99th percentile latency

Alert: Request latency > 1 second (p95)

Certificate Monitoring

TLS Certificate Status:

• Certificate expiration date
• Days until expiry
• Certificate validity

Alert: Certificate expires within 30 days

Cluster Health

Overall Status

Cluster Indicators:

• Total node count
• Healthy nodes vs total
• Total pod count
• Running pods vs total
• Resource utilization overview

Resource Capacity

Cluster Resources:

• Total CPU capacity
• Total memory capacity
• Total storage capacity
• Allocated vs available resources

Component Health

Control Plane:

• API server availability
• Controller manager status
• Scheduler status
• etcd health

Metrics Collection

Update Intervals

Metrics are collected and updated at a configurable interval (60 seconds by default):

Configure via ClusterPirate helm chart:

clusterPirate:
  metrics:
    updateIntervalSeconds: 60

Metric Retention

• Real-time metrics: Available immediately
• Historical metrics: Retained for 90 days (default)
• Cache TTL: 24 hours (86400 seconds)

Configure cache:

clusterPirate:
  metrics:
    cache:
      ttl: 86400

Viewing Metrics

Web Console

Access metrics through the portal:

1. Navigate to portal.cloudpirates.io
2. Select workspace and observability instance
3. Choose cluster
4. View metrics dashboard with real-time data

Dashboard Features:

• Interactive charts and graphs
• Time range selection
• Resource filtering
• Alert status indicators

API Reference

Metrics are exposed through the Kubernetes resource API endpoints.

See Kubernetes Resources for API details.

Related Resources

• Kubernetes Resources
• Events & Logs
• Alert Reference
• Setup Instructions