kubernetes

Abstract timeline visualization showing the seven-week evolution of an eBPF GPU agent from TCP-retransmit-inferred NCCL signals (v0.11.0, kernel-side) to a cluster-side MCP tool surface (v0.17.0)

eBPF, GPU Debugging, GPU Observability, MLOps

From TCP Retransmits to MCP-Driven Cluster Investigations: An eBPF GPU Agent Retrospective

Ingero Team / May 14, 2026

Seven weeks, ten releases. How an eBPF GPU agent grew an MCP tool surface that drives NCCL stall investigations from TCP retransmits to cluster scale.

Clock showing 3 AM next to a timeline of eBPF-traced events from pager fire to root cause found in 60 seconds: GPU stall, NFS checkpoint timeout identified

eBPF, GPU Debugging, GPU Observability, MLOps

GPU Incident at 3am: eBPF Tracing from Page to Root Cause in 60 Seconds

Ingero Team / March 19, 2026

3am page: GPU training pipeline missed its SLA. Datadog showed 95% GPU utilization, nvidia-smi agreed, everything looked green. Kernel-level tracing told a different story, and the root cause was visible in 60 seconds.