Components

• Node pool — 9 cases · node-pool
• Cluster autoscaler — 5 cases · cluster-autoscaler
• Kubernetes API server — 4 cases · kube-apiserver
• AWS VPC CNI — 4 cases · vpc-cni
• VPC subnet or pod address range — 3 cases · subnet
• Argo CD — 2 cases · argocd
• Configuration templating — 2 cases · config-templating
• CoreDNS — 2 cases · coredns
• Client-side DNS resolver and cache — 2 cases · dns-resolver
• GKE control plane — 2 cases · gke-control-plane
• GPU driver — 2 cases · gpu-driver
• HAProxy — 2 cases · haproxy
• InfiniBand fabric — 2 cases · infiniband
• Ingress controller — 2 cases · ingress
• kubelet — 2 cases · kubelet
• Container memory limit — 2 cases · memory-limit
• NCCL — 2 cases · nccl
• PyTorch — 2 cases · pytorch
• Admission webhook — 1 case · admission-webhook
• CI system — 1 case · ci-system
• conntrack table — 1 case · conntrack
• Container CPU limit — 1 case · cpu-limit
• Credential broker for pods — 1 case · credential-broker
• Application database — 1 case · database
• Descheduler — 1 case · descheduler
• etcd — 1 case · etcd
• Horizontal pod autoscaler — 1 case · hpa
• Karpenter — 1 case · karpenter
• kube-proxy — 1 case · kube-proxy
• Node launch template — 1 case · launch-template
• Linkerd — 1 case · linkerd
• Linux kernel — 1 case · linux-kernel
• Model hub — 1 case · model-hub
• Model request routing — 1 case · model-routing
• Node root disk — 1 case · node-disk
• Node operating system image — 1 case · node-image
• Persistent volumes and stateful sets — 1 case · persistent-volume
• Pod priority and preemption — 1 case · pod-priority
• Prometheus — 1 case · prometheus
• Container registry and its credentials — 1 case · registry
• Telemetry agent — 1 case · telemetry-agent
• TPU — 1 case · tpu
• XLA compiler — 1 case · xla