Explore

MCP servers and skills: search, filter by where they run, and compare.

1–3 of 3 tools for “pods”

gke-ai-troubleshooting-tpu-performance-degradationgoogle
Local
Diagnose GKE Cloud TPU training throughput drops and step-time regressions (15%+ TPU duty-cycle drop) using ML Diagnostics Workload Monitoring (`gcloud alpha mldiagnostics monitored-events` / `hyperco
in the skill textsaturation). Use when TPU training throughput or duty cycle drops without crashing pods, when `PERFORMANCE_DEGRADATION` monitored events fire, or when triaging slow multi-slice training
OfficialSkillApache-2.0★ 21k in collectionNo key