Q: Have you worked with Azure Kubernetes Service (AKS)? How would you deploy, monitor, and troubleshoot applications on AKS?
End-to-end operational guide for Azure Kubernetes Service (AKS): control plane vs node pools, Azure CNI vs Kubenet, GitOps deployment with Azure DevOps / GitHub Actions, and diagnosing pod failures via Azure Container Insights.
Want to master this scenario in a live sandbox? Stephane Maarek's AWS Certified DevOps Engineer Professional Masterclass on Udemy covers this exact problem with hands-on terminal drills.
🛠️ Production Runbook & Step-by-Step Resolution
Direct Answer: Yes, extensively. AKS provides a managed control plane while offloading worker node management into System and User node pools. Managing AKS effectively requires understanding Azure CNI networking, Workload Identity, and Azure Container Insights.
AKS Architecture & Networking Fundamentals
Control plane and networking choices:
- Architecture: Free/Standard tier managed control plane (etcd, API server managed by Microsoft); customer pays only for Virtual Machine Scale Set (VMSS) worker nodes grouped into System (CoreDNS, Metrics Server) and User (microservices) node pools.
- Azure CNI vs Kubenet: Kubenet uses internal pod overlay subnets (saves VNet IPs). Azure CNI gives every pod a real, routable IP from the Azure VNet subnet (lower latency, Direct VNet integration, but requires careful VNet CIDR sizing to avoid IP exhaustion).
Deploying Applications to AKS
Modern automated deployment workflow:
- CI pipeline builds image and pushes to Azure Container Registry (ACR).
- Authenticates to AKS using Microsoft Entra Workload Identity (OIDC federation—no long-lived service principal client secrets stored in CI).
- CD pipeline (or ArgoCD GitOps) renders Helm templates and applies manifests to AKS.
Monitoring & Troubleshooting in AKS
Diagnostic workflow when issues occur:
- Monitoring: Enable Azure Monitor Container Insights with Managed Prometheus and Grafana. Run KQL queries in Log Analytics:
ContainerInventory | where ContainerStatus == 'Failed'. - Troubleshooting Pods: Standard
kubectl describe podandkubectl logs --previous. - AKS Diagnose and Solve Problems: Native Azure Portal blade that runs automated diagnostic checks on node readiness, subnet IP allocation, and API server throttles.
- Node Issues: Check VMSS instance health in Azure Portal or run
az aks check-acrto verify network connectivity between AKS nodes and ACR.
- Architecture: Microsoft-managed control plane + VMSS worker node pools (System pool for core add-ons, User pool for workloads).
- Networking: Azure CNI gives pods native VNet IPs; requires large subnets to prevent IP exhaustion.
- Security: Entra Workload Identity federates Kubernetes ServiceAccounts with Azure Managed Identities (zero stored keys).
- Deployment: Azure Pipelines / GitHub Actions -> build & scan image -> push to ACR -> deploy via Helm / ArgoCD.
- Troubleshooting: Use 'kubectl describe/logs' for pod issues; 'az aks check-acr' for registry connectivity; and the Azure Portal 'Diagnose and Solve Problems' blade for node and network health.
Senior hiring managers and bar raisers evaluate how you prioritize during outages. Avoid these common anti-patterns:
- Engineering bespoke in-house developer tools instead of adopting standardized platform Golden Paths when addressing applications Bespoke internal tools create long-term maintenance debt and fragmented workflows, whereas standardized Golden Paths accelerate delivery.
- Overlooking infrastructure cost visibility and FinOps allocation tagging across multi-tenant environments Without granular resource tagging, engineering organizations cannot attribute cloud infrastructure costs to specific teams or business units.
- Failing to define explicit Service Level Objectives (SLOs) and Error Budgets to balance feature velocity with reliability Without agreed SLOs, engineering and product teams lack an objective data-driven framework to prioritize technical debt over new features.
- Neglecting automated continuous drift detection between declarative configuration and deployed cloud state Unmanaged drift accumulates over time, causing future automated deployments to fail or overwrite emergency manual adjustments.