Kubectl Skill

This skill enables comprehensive Kubernetes cluster management using kubectl and related tools.

Environment

Cluster Aliases

Three cluster/namespace combinations are pre-configured:

Usage:

k1 get pods          # 查看生产环境 pods
k2 get pods          # 查看预发布环境 pods
k get pods           # 查看本地环境 pods

Additional Tools

• kubectx – Switch between clusters
• kubens – Switch between namespaces
• argocd – GitOps deployments (see separate skill)
• kargo – Progressive delivery (see separate skill)

Safety Protocol

Dangerous Operations Requiring Confirmation

Before executing any of the following operations, explicitly confirm with the user:

• Delete operations: delete pod, delete deployment, delete service, delete pvc
• Scale to zero: scale --replicas=0
• Production modifications: Any k1 command that modifies resources
• Drain/cordon nodes: drain, cordon, uncordon
• Apply/patch: Changes to production resources

Confirmation Format

⚠️ 危险操作确认

环境: [Production/Staging/Local]
操作: [具体操作描述]
资源: [受影响的资源]
影响: [潜在影响说明]

是否继续执行？

Common Operations Reference

Resource Viewing

Pods

# List pods with status
k1 get pods
k1 get pods -o wide                    # Include node and IP info
k1 get pods --show-labels              # Show labels
k1 get pods -l app=simplex-api         # Filter by label

# Pod details
k1 describe pod <pod-name>

# Watch pods in real-time
k1 get pods -w

Deployments

# List deployments
k1 get deployments
k1 get deploy -o wide

# Deployment details
k1 describe deployment <name>

# Rollout status
k1 rollout status deployment/<name>

# Rollout history
k1 rollout history deployment/<name>

Services & Endpoints

# List services
k1 get services
k1 get svc

# Service details with endpoints
k1 describe svc <name>
k1 get endpoints <name>

All Resources

# Get all common resources
k1 get all

# Get specific resource types
k1 get pods,svc,deploy

# Get all resources with labels
k1 get all -l app=simplex-api

Logs & Debugging

Viewing Logs

# Basic logs
k1 logs <pod-name>

# Follow logs (streaming)
k1 logs -f <pod-name>

# Last N lines
k1 logs --tail=100 <pod-name>

# Logs since time
k1 logs --since=1h <pod-name>
k1 logs --since=10m <pod-name>

# Previous container logs (after restart)
k1 logs --previous <pod-name>

# Multi-container pod
k1 logs <pod-name> -c <container-name>

# All containers in pod
k1 logs <pod-name> --all-containers=true

Executing Commands

# Execute command in container
k1 exec <pod-name> -- <command>

# Interactive shell
k1 exec -it <pod-name> -- /bin/sh
k1 exec -it <pod-name> -- /bin/bash

# Specific container in multi-container pod
k1 exec -it <pod-name> -c <container> -- /bin/sh

Debugging

# Pod events and status
k1 describe pod <pod-name>

# Get pod YAML
k1 get pod <pod-name> -o yaml

# Debug with ephemeral container
k1 debug <pod-name> -it --image=busybox

# Check resource usage
k1 top pods
k1 top nodes

Deployment Management

Scaling

# Scale deployment
k1 scale deployment/<name> --replicas=3

# Autoscale
k1 autoscale deployment/<name> --min=2 --max=5 --cpu-percent=80

Rolling Updates

# Update image
k1 set image deployment/<name> <container>=<image>:<tag>

# Rollout status
k1 rollout status deployment/<name>

# Pause/resume rollout
k1 rollout pause deployment/<name>
k1 rollout resume deployment/<name>

# Rollback
k1 rollout undo deployment/<name>
k1 rollout undo deployment/<name> --to-revision=2

Restart

# Restart deployment (rolling restart)
k1 rollout restart deployment/<name>

Configuration Resources

ConfigMaps

# List ConfigMaps
k1 get configmaps
k1 get cm

# View ConfigMap content
k1 describe cm <name>
k1 get cm <name> -o yaml

# Create from file
k1 create configmap <name> --from-file=<path>

# Create from literal
k1 create configmap <name> --from-literal=key=value

Secrets

# List Secrets
k1 get secrets

# View Secret (base64 encoded)
k1 get secret <name> -o yaml

# Decode Secret value
k1 get secret <name> -o jsonpath='{.data.password}' | base64 -d

# Create Secret
k1 create secret generic <name> --from-literal=password=xxx

PersistentVolumeClaims

# List PVCs
k1 get pvc

# PVC details
k1 describe pvc <name>

Network Operations

Port Forwarding

# Forward local port to pod
k1 port-forward pod/<name> 8080:80

# Forward to service
k1 port-forward svc/<name> 8080:80

# Background port-forward
k1 port-forward pod/<name> 8080:80 &

Service Exposure

# Expose deployment as service
k1 expose deployment/<name> --port=80 --target-port=8080

# Get service external IP
k1 get svc <name> -o jsonpath='{.status.loadBalancer.ingress[0].hostname}'

Cluster Management

Nodes

# List nodes
k1 get nodes
k1 get nodes -o wide

# Node details
k1 describe node <name>

# Node resource usage
k1 top nodes

Namespaces

# List namespaces
k1 get namespaces

# Switch namespace (using kubens)
kubens <namespace>

# Create namespace
k1 create namespace <name>

Context Management

# List contexts
kubectx

# Switch context
kubectx <context-name>

# Show current context
kubectl config current-context

Resource Monitoring

# Pod resource usage
k1 top pods
k1 top pods --sort-by=cpu
k1 top pods --sort-by=memory

# Node resource usage
k1 top nodes

# HPA status
k1 get hpa
k1 describe hpa <name>

Output Formatting

For Status Checks

Provide concise summaries:

✅ Pod 状态 (production)
┌──────────────────────────┬─────────┬──────────┬─────────┐
│ Pod                      │ Status  │ Restarts │ Age     │
├──────────────────────────┼─────────┼──────────┼─────────┤
│ simplex-api-xxx-abc      │ Running │ 0        │ 2d      │
│ simplex-api-xxx-def      │ Running │ 0        │ 2d      │
└──────────────────────────┴─────────┴──────────┴─────────┘

For Troubleshooting

When investigating issues, gather:

1. Pod status: k1 get pod <name>
2. Pod events: k1 describe pod <name>
3. Recent logs: k1 logs --tail=50 <name>
4. Resource usage: k1 top pod <name>

Custom Output Formats

# JSON output
k1 get pods -o json

# YAML output
k1 get pod <name> -o yaml

# Custom columns
k1 get pods -o custom-columns=NAME:.metadata.name,STATUS:.status.phase

# JSONPath
k1 get pods -o jsonpath='{.items[*].metadata.name}'

Troubleshooting Workflows

Pod Not Starting

1. Check pod status: k1 get pod <name>
2. Check events: k1 describe pod <name> (look at Events section)
3. Check logs: k1 logs <name> or k1 logs --previous <name>
4. Common issues:
   • ImagePullBackOff: Check image name and registry credentials
   • CrashLoopBackOff: Check application logs
   • Pending: Check resource requests and node capacity

High Resource Usage

1. Check pod usage: k1 top pods --sort-by=memory
2. Check node usage: k1 top nodes
3. Check HPA status: k1 get hpa
4. Consider scaling: k1 scale deployment/<name> --replicas=N

Service Not Accessible

1. Check service: k1 get svc <name>
2. Check endpoints: k1 get endpoints <name>
3. Check pod labels match service selector
4. Test from within cluster: k1 exec -it <pod> -- curl <service>:<port>

Integration Notes

For GitOps operations (deployments via git), use the ArgoCD and Kargo skills:

• ArgoCD: Application sync, rollback, status
• Kargo: Progressive delivery, freight promotion

For AWS infrastructure operations, use the AWS CLI skill.