Guard: Scalable Straggler Detection and Node Health Management for Large-Scale Training

TL;DR

Guard is a scalable system that detects stragglers and manages node health in large-scale GPU training, improving efficiency and stability during multi-month foundation model pretraining.

Contribution

It introduces a novel combination of online performance monitoring and offline node qualification to detect fail-slow behaviors and failures in large-scale training clusters.

Findings

Up to 1.7x increase in FLOPs utilization

Reduced training step variance from 20% to 1%

Increased mean time to failure (MTTF)

Abstract

Training frontier-scale foundation models involves coordinating tens of thousands of GPUs over multi-month runs, where even minor performance degradations can accumulate into substantial efficiency losses. Existing health-check mechanisms, such as NCCL tests or GPU burn-in, primarily focus on functional correctness and often fail to detect fail-slow behaviors that silently degrade system performance. In this paper, we present Guard, a scalable system for detecting stragglers and ensuring node health in large-scale training clusters. Guard combines lightweight online performance monitoring during training with an offline node-sweep mechanism that systematically evaluates and qualifies nodes before they participate in production workloads. This design enables Guard to detect both acute failures and long-running fail-slow behaviors that traditional diagnostics cannot capture. Deployed on large-scale foundation model pretraining workloads, Guard improves mean FLOPs utilization by up to 1.7x, reduces run-to-run training step variance from 20% to 1%, increases mean time to failure (MTTF), and significantly reduces operational and debugging overhead. These results demonstrate that proactive straggler detection and systematic node qualification are critical for maintaining stable and efficient large-scale training.