#Pod Isolation

#Introduction

Pod Isolation enables dynamic exclusion of malfunctioning Pods from service traffic while maintaining their operational state. This functionality:

• Prevents failed requests to defective endpoints
• Maintains service availability during troubleshooting
• Allows seamless reintegration post-recovery

Core value: Ensures continuous service reliability through targeted traffic management

#Features

• Single-click Pod exclusion/inclusion
• Real-time health metrics integration
• Legacy service compatibility mode
• Non-destructive isolation (Pod remains running)

#Advantages

Precision: Target specific Pods without service restart Safety: Maintains quorum for remaining Pods Visibility: Integrated success rate/latency tracking Compliance: Works within Kubernetes orchestration

#Isolating Pods

#Step 1: Access Pod Management

1. Navigate: Service List > Target Service
2. Select Pod Group tab
3. Locate target Pod via metrics or logs

#Step 2: Execute Isolation

# Isolation state annotation
metadata:
  annotations:
    asm.cpaas.io/isolated: "true"

UI Control	Action
	Open context menu
Set Isolation	Toggle isolation state
Set	Confirm configuration

#Legacy Service Activation

#Requirement Checklist

• Kubernetes v1.18+
• Service created prior to isolation feature release
• Cluster admin privileges

#Configuration Update

1. Modify Deployment labels:

metadata:
  labels:
    asm.cpaas.io/msselector: product-service

2. Update internal route selector:

spec:
  selector:
    asm.cpaas.io/msselector: product-service

#Monitoring & Validation

#Key Metrics Dashboard

Metric	Threshold	Refresh Control
Error Rate	<5%
Latency	<500ms
Active Connections	±10% variance

#Status Verification

kubectl get endpoints <service-name> -o jsonpath='{.subsets[].notReadyAddresses}'

#Operational Constraints

1. State Limitations:

   • Only affects Pods in Running state
   • Excludes Pods in Completed/CrashLoopBackOff

2. Conflict Scenarios:

   • Disabled during canary releases
   • Overridden by scale-in/scale-out events

3. Persistence:

   • Isolation resets on Pod restart
   • Manual re-isolation required after recovery