Issues

• Persistent async checkpointing retains GPU sources after D2H staging

  NVIDIA/nvidia-resiliency-ext#433 · Filter by author Elcfin opened on Oct 5, 2026

• CPU-shm AsyncCallsQueue imports during interpreter shutdown after explicit close

  NVIDIA/nvidia-resiliency-ext#423 · Filter by author bzantium opened on Sep 11, 2026

• performance improvements during cpu-shm snapshotting

  NVIDIA/nvidia-resiliency-ext#405 · Filter by author ankurvdev opened on Aug 14, 2026
  1

• [DOCS] Fix invalid module import path in Straggler Detection documentation (nvidia_resiliency_ext.attribution)

  NVIDIA/nvidia-resiliency-ext#384 · Filter by author Faisal-Alsrheed opened on Jul 31, 2026

• Request Python 3.13 and 3.14 wheel support for nvidia-resiliency-ext

  NVIDIA/nvidia-resiliency-ext#366 · Filter by author LitLeo opened on Jul 1, 2026
  1

• Local checkpointing replication writes NaN-corrupted replica shards for TE/MXFP8 quantized weights (0.5.0) — primaries bitwise-clean; root cause of Megatron-LM#5281

  NVIDIA/nvidia-resiliency-ext#349 · Filter by author cxiong-ship-it opened on Jun 11, 2026
  1

• ft_launcher 0.6.0 hot-spare: re-rendezvous with live agents > max_nodes permanently closes the rendezvous ('Maximum number of nodes exceeded') and kills the whole job

  NVIDIA/nvidia-resiliency-ext#348 · Filter by author cxiong-ship-it opened on Jun 11, 2026
  2

  

• logsage hard dependency causes transitive dependency conflicts for downstream consumers

  NVIDIA/nvidia-resiliency-ext#301 · Filter by author terrykong opened on Apr 21, 2026
  1

  

• Straggler Detector: add process_group parameter for DTensor/Ray compatibility

  NVIDIA/nvidia-resiliency-ext#277 · Filter by author dmvevents opened on Mar 16, 2026
  4

• Wheels for more python versions and cu13

  NVIDIA/nvidia-resiliency-ext#234 · Filter by author sakgoyal opened on Dec 10, 2025

• Questions about the ft_launcher process

  NVIDIA/nvidia-resiliency-ext#182 · Filter by author zhougf-zhiAn opened on Sep 11, 2025
  2