Cluster Scaling

WuKongIM Kubernetes cluster scaling operations guide

WuKongIM supports dynamic scaling in Kubernetes environments, allowing flexible adjustment of cluster size based on business requirements.

Description

Applicable scenarios: Applications with high data security requirements, large applications.

Advantages: High availability, strong disaster recovery, supports online scaling, real-time automatic backup between multiple replicas, load balancing, no manual configuration required, fast scaling.

Note: WuKongIM currently supports hot scaling but does not support hot shrinking! Does not support hot shrinking! Does not support hot shrinking!

Scaling Operations

Scale Up Cluster

# Scale to 3 replicas
helm upgrade wkim wukongim/wukongim \
  -n wukongim \
  --create-namespace \
  --version 0.1.0 \
  --set replicaCount=3

Optional parameters:

• replicaCount=3: Number of replicas (default is 2)
• externalIP=<IP_ADDRESS>: External IP address

Scale to Different Sizes

# Scale to 5 replicas for high load
helm upgrade wkim wukongim/wukongim \
  -n wukongim \
  --version 0.1.0 \
  --set replicaCount=5

# Scale to 7 replicas for very high load
helm upgrade wkim wukongim/wukongim \
  -n wukongim \
  --version 0.1.0 \
  --set replicaCount=7

Verify Scaling

# Check Pod status
kubectl get pods -n wukongim

# Check cluster nodes
kubectl port-forward svc/wkim-wukongim 5001:5001 -n wukongim &
curl http://localhost:5001/cluster/nodes

# View detailed cluster information
curl http://localhost:5001/cluster/nodes | jq '.'

Scaling Best Practices

Pre-Scaling Checklist

1. Resource Planning: Ensure Kubernetes cluster has sufficient resources
2. Monitoring: Check current resource usage and performance metrics
3. Backup: Create backup before scaling operations
4. Load Testing: Verify current performance baseline

Recommended Scaling Patterns

For Growing Load:

# Start with 3 replicas (minimum for HA)
helm upgrade wkim wukongim/wukongim -n wukongim --set replicaCount=3

# Scale to 5 for moderate load increase
helm upgrade wkim wukongim/wukongim -n wukongim --set replicaCount=5

# Scale to 7 for high load
helm upgrade wkim wukongim/wukongim -n wukongim --set replicaCount=7

Resource Allocation per Replica:

# Scale with resource adjustments
helm upgrade wkim wukongim/wukongim \
  -n wukongim \
  --set replicaCount=5 \
  --set resources.requests.memory=4Gi \
  --set resources.requests.cpu=2000m \
  --set resources.limits.memory=8Gi \
  --set resources.limits.cpu=4000m

Post-Scaling Verification

# Check all pods are running
kubectl get pods -n wukongim -o wide

# Verify cluster health
kubectl exec -n wukongim deployment/wkim-wukongim -- \
  curl -f http://localhost:5001/health

# Check cluster status
kubectl port-forward svc/wkim-wukongim 5001:5001 -n wukongim &
curl http://localhost:5001/cluster/nodes

# Monitor resource usage
kubectl top pods -n wukongim
kubectl top nodes

Monitoring During Scaling

Watch Scaling Progress

# Watch pods being created
kubectl get pods -n wukongim -w

# Monitor events
kubectl get events -n wukongim --sort-by='.lastTimestamp'

# Check deployment status
kubectl rollout status deployment/wkim-wukongim -n wukongim

Performance Monitoring

# Check resource usage
kubectl top pods -n wukongim

# View logs during scaling
kubectl logs -f deployment/wkim-wukongim -n wukongim

# Monitor cluster metrics
kubectl port-forward svc/wkim-wukongim 5300:5300 -n wukongim &
curl http://localhost:5300/metrics

Troubleshooting Scaling Issues

Common Scaling Problems

Insufficient Resources:

# Check node resources
kubectl describe nodes

# Check resource requests vs limits
kubectl describe deployment wkim-wukongim -n wukongim

# View pod events
kubectl describe pods -n wukongim

Pod Startup Issues:

# Check pod status
kubectl get pods -n wukongim

# View pod logs
kubectl logs <pod-name> -n wukongim

# Check pod events
kubectl describe pod <pod-name> -n wukongim

Cluster Communication Issues:

# Test inter-pod communication
kubectl exec -it <pod-name> -n wukongim -- ping <other-pod-ip>

# Check cluster ports
kubectl exec -it <pod-name> -n wukongim -- netstat -tulpn | grep 11110

# View cluster logs
kubectl logs <pod-name> -n wukongim | grep cluster

Recovery Procedures

# Rollback to previous replica count if issues occur
helm rollback wkim -n wukongim

# Force restart problematic pods
kubectl delete pod <pod-name> -n wukongim

# Check and fix resource constraints
kubectl patch deployment wkim-wukongim -n wukongim -p '{"spec":{"template":{"spec":{"containers":[{"name":"wukongim","resources":{"requests":{"memory":"4Gi","cpu":"2000m"}}}]}}}}'

Automated Scaling

Horizontal Pod Autoscaler (HPA)

# hpa.yaml
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: wkim-hpa
  namespace: wukongim
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: wkim-wukongim
  minReplicas: 3
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 70
  - type: Resource
    resource:
      name: memory
      target:
        type: Utilization
        averageUtilization: 80

Apply HPA:

kubectl apply -f hpa.yaml
kubectl get hpa -n wukongim

Vertical Pod Autoscaler (VPA)

# vpa.yaml
apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: wkim-vpa
  namespace: wukongim
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: wkim-wukongim
  updatePolicy:
    updateMode: "Auto"
  resourcePolicy:
    containerPolicies:
    - containerName: wukongim
      maxAllowed:
        cpu: 4
        memory: 8Gi
      minAllowed:
        cpu: 500m
        memory: 1Gi

Performance Considerations

Optimal Replica Counts

Load Level	Recommended Replicas	Use Case
Light	3	Development, small teams
Medium	5	Growing applications
High	7	Large applications
Very High	9+	Enterprise scale

Resource Planning

# Calculate total resources needed
# For 5 replicas with 4Gi memory each = 20Gi total memory required
# For 5 replicas with 2 CPU each = 10 CPU cores total required

helm upgrade wkim wukongim/wukongim \
  -n wukongim \
  --set replicaCount=5 \
  --set resources.requests.memory=4Gi \
  --set resources.requests.cpu=2000m

Next Steps

Multi-Node Deployment

Learn about multi-node cluster deployment

Upgrade Guide

Upgrade WuKongIM in Kubernetes

Monitoring Setup

Set up monitoring for scaled cluster

Performance Tuning

Optimize performance for scaled deployment