# Warning FailedCreatePodSandBox: Failed to create pod sandbox: rpc error: code = Unknown desc = failed to set up sandbox container "" network for pod "my-pod": networkPlugin cni failed to set up pod "my-pod_default" network: no IP addresses available in network

- **ID:** `cloud/azure-aks-pod-creation-pending-no-ip`
- **Domain:** cloud
- **Category:** resource_error
- **Error Code:** `FailedCreatePodSandBox`
- **Verification:** ai_generated
- **Fix Rate:** 85%

## Root Cause

Azure Kubernetes Service (AKS) node pool has exhausted its subnet IP address range, preventing new pods from being assigned IP addresses.

## Version Compatibility

| Version | Status | Introduced | Deprecated |
|---------|--------|------------|------------|
| AKS 1.26 | active | — | — |
| AKS 1.27 | active | — | — |
| Azure CNI | active | — | — |

## Workarounds

1. **Increase the subnet size by adding a new subnet and updating the AKS cluster to use it. First, create a new subnet: `az network vnet subnet create --resource-group my-rg --vnet-name my-vnet --name new-subnet --address-prefixes 10.0.1.0/24` Then update the node pool: `az aks nodepool update --resource-group my-rg --cluster-name my-cluster --name my-nodepool --vnet-subnet-id /subscriptions/.../subnets/new-subnet`** (90% success)
   ```
   Increase the subnet size by adding a new subnet and updating the AKS cluster to use it. First, create a new subnet: `az network vnet subnet create --resource-group my-rg --vnet-name my-vnet --name new-subnet --address-prefixes 10.0.1.0/24` Then update the node pool: `az aks nodepool update --resource-group my-rg --cluster-name my-cluster --name my-nodepool --vnet-subnet-id /subscriptions/.../subnets/new-subnet`
   ```
2. **Delete unused pods to free IPs. Use `kubectl delete pod <pod-name>` for non-critical pods. Then monitor with `kubectl get pods -o wide` to see IP reclamation.** (70% success)
   ```
   Delete unused pods to free IPs. Use `kubectl delete pod <pod-name>` for non-critical pods. Then monitor with `kubectl get pods -o wide` to see IP reclamation.
   ```

## Dead Ends

- **Restart the AKS cluster** — Restarting does not free up IP addresses; the subnet is still full. (95% fail)
- **Scale down the node pool to reduce node count** — Scaling down reduces nodes but does not reclaim IPs from existing pods. Pod IPs remain allocated until pods are deleted. (80% fail)