Architecture Overview¶

Good AKS troubleshooting starts by locating the failure layer: Kubernetes object, node, cluster integration, or Azure dependency.

Main Content¶

flowchart TD
    U[Users] --> G[Ingress / Gateway]
    G --> S[Service]
    S --> P[Pod]
    P --> N[Node]
    N --> C[Cluster Control Plane]
    P --> X[External Dependencies]
    N --> A[Azure Network and Storage]

Failure layers¶

• Pod layer: image pulls, crashes, readiness, configuration.
• Service/Ingress layer: selector mismatch, endpoints missing, TLS or routing issues.
• Node layer: NotReady status, pressure, daemonset failure, IP exhaustion.
• Azure integration layer: load balancer, managed disk, identity, registry, or DNS issues.

Why this matters¶

Symptoms often appear at one layer and originate in another. A 502 at ingress might be a pod readiness issue, a service selector issue, or a node networking issue.

See Also¶

• Decision Tree
• Evidence Map
• Mental Model
• Platform: Cluster Architecture

Sources¶

• Troubleshoot AKS clusters
• AKS troubleshooting articles
• AKS core concepts for Kubernetes and workloads
• Azure Kubernetes Service (AKS) architecture