Performance Co-Pilot Helm Chart

A Helm chart for deploying Performance Co-Pilot (PCP) on Kubernetes.

Performance Co-Pilot is a system performance analysis toolkit that provides comprehensive real-time and historical metrics collection and analysis capabilities. PCP includes:

• Real-time monitoring: Live system metrics via pmcd daemon
• Historical analysis: Archive logging via pmlogger
• Web interface: REST API via pmproxy for integration with Grafana, Prometheus
• Extensive metrics: CPU, memory, disk, network, containers, and custom application metrics
• eBPF support: Advanced kernel tracing with bpftrace and BPF agents
• Multi-host monitoring: Central collection from distributed systems

For more information, visit pcp.io or the GitHub repository.

Pre-built packages are available at packagecloud.io/performancecopilot/pcp.

Prerequisites

• Kubernetes 1.19+
• Helm 3.0+
• PV provisioner support in the underlying infrastructure (for persistent storage)

Installing the Chart

helm install pcp ./pcp

Uninstalling the Chart

helm uninstall pcp

Configuration

Parameter	Description	Default
replicaCount	Number of replicas (fixed at 1)	1
image.repository	PCP container image repository	quay.io/performancecopilot/pcp
image.tag	PCP container image tag	latest
image.pullPolicy	Image pull policy	IfNotPresent
hostNetwork	Use host networking	false
securityContext.privileged	Run in privileged mode	true
persistence.pmlogger.enabled	Enable persistent storage for pmlogger	true
persistence.pmlogger.size	Size of pmlogger volume	1Gi
persistence.pmproxy.enabled	Enable persistent storage for pmproxy	true
persistence.pmproxy.size	Size of pmproxy volume	1Gi
hostMonitoring.enabled	Enable host monitoring mode	false
env.PCP_SERVICES	PCP services to start	pmcd,pmie,pmlogger,pmproxy
env.PCP_DOMAIN_AGENTS	PCP agents to enable	""
env.HOST_MOUNT	Host mount path for monitoring	""
env.KEY_SERVERS	Redis/Valkey connection specs	localhost:6379

Host Monitoring

For comprehensive system monitoring including eBPF capabilities:

helm install pcp ./pcp \
  --set hostMonitoring.enabled=true \
  --set hostNetwork=true \
  --set env.HOST_MOUNT="/host" \
  --set env.PCP_DOMAIN_AGENTS="bpf,bpftrace"

Accessing PCP

• PMCD: Port 44321 (native PCP tools)
• PMPROXY: Port 44322 (HTTP/REST API)

Examples

Basic deployment:

helm install pcp ./pcp

With custom agents:

helm install pcp ./pcp --set env.PCP_DOMAIN_AGENTS="postgresql,apache"

Host monitoring with eBPF:

helm install pcp ./pcp \
  --set hostMonitoring.enabled=true \
  --set hostNetwork=true \
  --set env.HOST_MOUNT="/host" \
  --set env.PCP_DOMAIN_AGENTS="bpf,bpftrace"

Notes

• PCP runs as a single instance (replicaCount=1) since multiple replicas would duplicate system monitoring data
• Privileged mode is required for full system access
• Host monitoring mode requires additional cluster permissions