Cast.ai insights

AWS ENI Limits Constrain Max Pod Density on Smaller Instance Typeswarning

AWS CNI enforces low max pod counts (e.g., 58 for c6g.2xlarge) on instances with fewer CPUs due to ENI limitations, preventing full utilization of node compute capacity unless prefix delegation is enabled.

AWS Instance Profile Role Removed from aws-auth Breaks Node Registrationcritical

When all EKS-managed node groups using a shared instance profile are deleted, AWS removes the instance profile role from aws-auth ConfigMap, breaking Cast AI-managed nodes that rely on the same role for kubelet authentication to the API server.

Node Drain Failure Due to Pod Disruption Budget Violationwarning

Cast AI attempts to drain nodes for optimization but fails when Pod Disruption Budgets (PDBs) with percentage-based unavailability limits prevent pod eviction, causing nodes to revert to active state.

Watchdog Emergency Capacity Activation During Cast AI Outagewarning

When Cast AI control plane becomes unavailable, Watchdog fallback mechanism detects the outage and provisions emergency compute capacity through cloud provider APIs, bypassing Cast AI's optimization logic.

Metrics Server Unavailability Breaks Workload Efficiency Monitoringwarning

Cast AI agent cannot discover or poll metrics from the Kubernetes Metrics Server, preventing workload efficiency analysis and rightsizing recommendations.

Topology Spread Constraint Violation Prevents Pod Schedulingcritical

Pods with topology spread constraints (e.g., zone distribution) remain unscheduled when Cast AI autoscaler detects it cannot satisfy zone requirements, either due to misconfigured availability zones or conflicting node affinity rules.

Cast AI Agent Connection Failure Blocks Cluster Discoverycritical

The Cast AI agent fails to authenticate or connect to cloud provider APIs (AWS EC2, GCP metadata) or the Cast AI control plane, preventing cluster registration and causing clusters to appear as 'Not responding' or 'Disconnected'.