Introduction
Pod Isolation Introduction
Pod Isolation enables immediate traffic cutoff to malfunctioning Pods while maintaining their operational state. This feature:
- Prevents request failures from defective Pods
- Maintains service availability during troubleshooting
- Allows seamless reintegration after repair
Core purpose: Ensure service continuity by dynamically managing endpoint participation
Advantages
-
Quick Isolation Single-click removal from service mesh without pod termination
-
Stability Preservation Maintains optimal success rates by excluding faulty instances
-
K8s Native Integration Works with standard Deployment resources
-
Observability Integrated with monitoring dashboards for health validation
Scenarios
-
Abnormal Pod Handling Isolate Pods with elevated 5xx errors or latency spikes
-
Maintenance Windows Temporarily remove nodes during infrastructure updates
-
Performance Debugging Test fixes on isolated Pod without affecting production traffic
-
Rolling Recovery Gradually reintroduce repaired Pods after validation
Limitations
-
State Constraints Only affects Running Pods, excludes Completed/CrashLoopBackOff
-
Version Compatibility Legacy services require label updates (asm.cpaas.io/msselector)
-
Conflict Scenarios Disabled during canary releases and horizontal scaling events
-
Persistence Isolation resets on pod restart or scale-down operations