Cluster autoscaler sees unschedulable pods but never scales
Pending pods are real, yet autoscaler takes no action because it cannot associate the node group with the cluster through the expected tag contract.
K8sAdvanced
Cluster autoscaler sees pending pods but still refuses to scale
Pending workloads are real, yet the autoscaler cannot prove a new node would help because the node-group template metadata is incomplete.
K8sAdvanced
An EKS add-on update looks clean but new pods cannot get IPs
The CNI daemonset is ready everywhere, yet density collapses on new nodes because their bootstrap path never enabled the IP mode the old nodes used.
K8sIntermediate
A DaemonSet never reaches the last node
Most nodes run the agent, yet one entire group stays empty because those nodes never received the identifying label the workload expects for placement.