Home/Refreshers/Containers on AKS and OKE: refresher
Containers on AKS and OKE
What each service records about a cluster, what a pod can reach on its node, and how to keep the node's disks.
| AKS | OKE | |
|---|---|---|
| Audit log | Off until a diagnostic setting; kube-audit-admin drops get and list | Changes in OCI Audit, 365 days |
| Keep a node | Autoscaler and do-not-disrupt annotations, cordon | Autoscaler annotation, cordon, pause pool changes |
| Node disks | Ephemeral OS disks cannot be snapshotted | Boot and block volume backups |
| Node identity from pods | IMDS reachable by default | Instance principal reachable by default |
Commands
| AKS: a node's VM | kubectl get node <node> -o jsonpath='{.spec.providerID}' |
| AKS: snapshot a disk elsewhere | az snapshot create -g <evidence group> -n <name> --source <disk id> --incremental true |
| OKE: a kubeconfig | oci ce cluster create-kubeconfig --cluster-id <ocid> --file <path> --token-version 2.0.0 |
| OKE: back up the boot volume | oci bv boot-volume-backup create --boot-volume-id <ocid> --type FULL |
Watch out
- AKS reboots, then reimages, a node NotReady for about 10 minutes. Do not cut the node off.
- kubectl 1.31 and later sends exec as a get, which kube-audit-admin leaves out.
- OCI has no termination protection. Node cycling replaces nodes.
Check it worked
A test exec appears in AKSAudit, and a captured node's disks have snapshots or backups tagged with the incident.