Solutions

Articles

• Getting Started with SLURM on Crusoe Cloud
• Slurm jobs fail with "ImportError: libGL.so.1" after node restarts
• Slurm login node restarts with OOMKilled (exit code 137)
• API key fails with "401 Authentication failed" after user removal
• Move Ethernet NIC Interrupts Off Your Training CPUs in Crusoe GPU VMs
• GPU Operator pods stuck on B200/B300 CMK nodes due to MOFED race condition
• Replacement instance inherits stale RAID metadata from a deleted instance
• Pods stuck in ContainerCreating or Terminating on NFS-backed PVCs
• VM Hangs With Kernel Soft Lockup in the NFS Write Path on Shared Storage
• Fixing pod scheduling failures from Cilium PodCIDR exhaustion
• High I/O Wait and Pressure Stall on a VM Accessing a Shared Disk
• B200 VM Unreachable After Reboot Due to Unsupported Ubuntu 22.04 Driver Stack
• NIXL Memory Registration Errors on InfiniBand-Enabled GPU Instances
• Prolog Script Updates Are Not Taking Effect — Nodes Drain After Job Submission
• CMK Cluster Autoscaler: Best practices and Known Limitations
• How-To configure Crusoe Managed Kubernetes as an OIDC provider so that workloads can access your public cloud resources
• GPU Operator CrashLoopBackOff Due to Exhausted inotify Limits
• GPU Operator Pod Creation Blocked by Orphaned Run:ai Volcano Webhook Configurations
• CMK Worker Node Boots Into Emergency Mode After Stop/Start Due to NVMe UUID Mismatch
• Cluster Autoscaler Conflicts When Scaling Down a CMK Nodepool
• PyTorch NCCL DNS Resolution Failures During Distributed Training Initialization
• Slurm Nodes in PLND State: Why Jobs Aren’t Starting and How to Fix It
• Get started with NVIDIA Riva Server on Crusoe Managed Kubernetes
• Resolving MOFED Storage Module Race Condition on CMK GPU Nodes
• Kubernetes Service Unreachable Using Tailscale Operator
• NCCL Hangs and Multi-Node Training Stalls Caused by Failed nvidia-fabricmanager
• Resolving crashes on NCCL initialization with GB200 NVL72 Slurm Cluster
• CUDA Validation Failure on B200 Nodes in CMK
• Understanding GB200 Performance Impact of Memory Spillover in NUMA Mode
• Nvidia GPU Operator-enabled Workloads Show in the Logs: "NVIDIA peer memory driver not detected" and "CUDA Forward Compatibility mode ENABLED"