# Monitoring Kubernetes

The below areas are components and metrics you should monitor for in a self-deployed Kubernetes cluster (AKS-Engine, Kops, Kubespray,etc...)

## Master Components

### etcd
- Up/Down of Service
- RPC Rate
- Request Latency
- etcd Memory usage
- Client and Peer Traffic In
- Client and Peer Traffic Out
- Disk Sync Duration
- Total Leader elections over 24hrs

### API
- Total API Request
- Dropped API Request
- API Request Latency
- HTTP Client Request
- Up/Down of Service

### Scheduler
- Total HTTP CLient Request
- HTTP Client Request Duration
- Scheduling Latency

### Controller Manager
- Healthy Nodes
- Unhealthy Nodes
- HTTP Client Request
- Request Duration

### CoreDNS
- Total Request
- Request by type
- Responses by type
- Response Duration

## Cluster Monitoring
- CPU Utilization
- Memory Utilization
- CPU Request
- CPU Limits
- Memory Request
- Memory Limits
- CPU/Memory Limits+Request per namespace

### Node Monitoring
- System Load
- Usage Per Core
- CPU Utilization
- Memory Utilization
- Disk I/O
- Disk Space
- Network Tramsmission