NeVibeCode · 10 items
Skills and MCP tagged “tpu”
gke-ai-troubleshooting-handle-disruption-gpu-tpu
official skill Diagnoses GKE node disruptions during Compute Engine maintenance for GPU and TPU workloads, then points to practical protection settings.
Claude Code Gemini CLI CX Codex
4,700installs
gke-ai-troubleshooting-tpu-vbar-oom
official skill Investigate vbar_control_agent crashes and OOMs on GKE TPU v6e nodes, especially around device resets or frequent metrics polling.
Claude Code Gemini CLI CX Codex
4,400installs
gke-ai-troubleshooting-tpu-metrics-monitoring
official skill Monitors GKE TPU workloads with system metrics and PromQL. Use it to investigate node health, TPU memory, node pool availability, and interruptions.
Claude Code Gemini CLI CX Codex
3,700installs
gke-ai-troubleshooting-tpu-dynamic-slices-monitoring
official skill Checks GKE TPU Dynamic Slice status, investigates provisioning failures, and validates manifests for single-slice and JobSet workloads.
Claude Code Gemini CLI CX Codex
3,700installs
gke-ai-troubleshooting-node-unresponsive-timeout
official skill Diagnoses unresponsive GKE TPU and GPU nodes stuck in NodeStatusUnknown. Pinpoints kernel panics and missing kubelet heartbeats when node auto-repair is disabled.
Claude Code Gemini CLI CX Codex
810installs
gke-ai-troubleshooting-tpu-performance-degradation
official skill This skill investigates Cloud TPU training slowdowns in GKE and separates hardware or network limits from workload resource bottlenecks. Use it when jobs keep running but training steps take longer.
Claude Code Gemini CLI CX Codex
691installs
gke-ai-troubleshooting-tpu-slice-mig-stuck-updating
official skill Diagnoses TPU Dynamic Slice creation failures and pending JobSets when the backing MIG is unstable. Use it for GKE issues tied to MIG updates or repairs.
Claude Code Gemini CLI CX Codex
440installs
gke-ai-troubleshooting-tpu-vm-repair-stuck
official skill Diagnoses GKE TPU VMs stuck in REPAIRING after host errors or maintenance. Use it when a TPU slice cannot be scheduled because a VM remains in repair.
Claude Code Gemini CLI CX Codex
431installs
gke-ai-troubleshooting-tpu-partition-overlap
official skill Diagnoses GKE TPU slice creation failures (SliceCreationFailed) caused by partition overlap with another slice or JobSet. Pinpoints node conflicts using read-only checks.
Claude Code Gemini CLI CX Codex
378installs
gke-ai-troubleshooting-tpu-reservation-stockout
official skill Diagnoses GKE TPU node pool creation failures caused by reservation stockouts when unhealthy nodes lack auto-repair and hoard reserved capacity.
Claude Code Gemini CLI CX Codex
377installs