Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- Use a VPC-native cluster at 1.35.3-gke.1155000 or later and node pools at 1.35.1-gke.1516000 or later. If you use Cloud Service Mesh or open-source Istio, use Istio 1.23 or later.
- Check Active Assist, existing network policies, Pod IP capacity, and maintenance windows. Google says the migration ignores the cluster's disruption budget.
- The cluster shows MIGRATE_TO_ADVANCED_DATAPATH during the transition. GKE finalizes automatically after all node pools migrate, and the cluster shows ADVANCED_DATAPATH.
- The rollback window closes when finalization completes. A finalized cluster cannot return to Dataplane V1 through this migration path.
Check which clusters can migrate in place
The preview covers eligible VPC-native clusters. The cluster must run at least 1.35.3-gke.1155000 and the node pools at least 1.35.1-gke.1516000. If you use Cloud Service Mesh or open-source Istio, check that Istio is at least 1.23.
Routes-based clusters are ineligible. A custom CNI plugin installed in place of GKE's managed plugin and custom eBPF programs attached to managed interfaces are also unsupported. Review Active Assist or the cluster details for migration conflicts before scheduling the change.
Assess network policies, Pod IPs, and disruption first
Google's preparation guide requires handling existing NetworkPolicy resources before migration. GKE Standard clusters with Calico network policy enabled cannot migrate directly. Inventory the services those policies protect, then verify isolation and connectivity on a non-production cluster before scheduling a production change.
GKE reschedules Pods when it recreates nodes. Check whether your upgrade strategy needs to run old and new nodes together and whether the Pod IP range can accommodate them. Google says the migration ignores the cluster's disruption budget. Schedule a low-traffic window and check the health signals for critical services.
Read datapathProvider for both the cluster and node pools
The cluster shows MIGRATE_TO_ADVANCED_DATAPATH during the transition. Check datapathProvider on individual node pools to see which ones have switched. On Autopilot, or Standard with Autopilot ComputeClasses, the control plane manages gk3- node pools. Google's guide says they cannot be listed or patched using the same node-pool steps.
If progress stalls, check maintenance windows, exclusions, and existing node-pool operations. For GCE_STOCKOUT or ZONE_RESOURCE_POOL_EXHAUSTED_WITH_DETAILS, check resource availability in that zone. Google says GKE automatically retries the affected node pools.
Decide whether to continue or roll back before finalization
Use representative workloads to check DNS, service connectivity, and policy enforcement before widening the migration. GKE finalizes automatically after all node pools migrate, changing the cluster status to ADVANCED_DATAPATH. Make the go or rollback decision before the last batch of node pools completes so there is time to respond.
Before finalization completes, use Google's documented rollback procedure to return to Dataplane V1. Track node-pool recovery and confirm final completion when the cluster shows LEGACY_DATAPATH. Once migration finalization completes, the documented in-place rollback path is unavailable.
Sources
Frequently Asked Questions
Can Terraform start this preview migration?
Google's migration guide excludes Terraform as a way to start this in-place migration. Follow the documented migration procedure and manage existing infrastructure configuration against the actual cluster state.
Can I turn on NodeLocal DNSCache during the migration?
Cluster-wide networking configuration is locked until migration completes. Schedule NodeLocal DNSCache and similar features before starting or after finalization. Configure Dataplane V2 observability after migration.
Does migration to Dataplane V2 also enable Ambient Networking?
Dataplane V2 is a prerequisite for Ambient Networking. Check Ambient Networking's own version, region, and cluster eligibility, then enable it separately if appropriate.