DevOps & Infrastructure
Gke AI Troubleshooting Tpu Mxla Hang
The skill diagnoses multi‑slice GKE Cloud TPU training hangs by analyzing Megascale XLA hang logs and Cloud Monitoring latency metrics. It distinguishes compiler or launch divergence from host data stalls and TPU, SparseCore, ICI, or network faults, enabling engineers to pinpoint the root cause of frozen training jobs that emit HANG_DETECTED. This service is ideal for teams experiencing abrupt training stalls, but not for gradual throughput drops or pod preemption scenarios.
- Publisher / source
- Google Skills (Verified)
- Upstream record
- google/skills
- Version
- Version 1
- Published
- Publication review
- Approved for publication · recorded passed scan 2026-10-09
- Content fingerprint
1a467aa00d8a1447c8c6b03fbc8b01e5cdcd4a40b436a145de07bacc0fdeeb8a