nvidia-device-plugin Failed to initialize NVML libnvidia-ml.so.1 cannot be found

The NVIDIA device plugin could not load libnvidia-ml.so, so it never registered nvidia.com/gpu with the kubelet. Every GPU pod on the node stays Pending with "Insufficient nvidia.com/gpu" while the node itself looks healthy. This entry explains how to confirm the cause, apply the fix, and separate it from adjacent containers failures.

The literal signature is "nvidia-device-plugin Failed to initialize NVML libnvidia-ml.so.1 cannot be found". It is a environment failure associated with NVIDIA Container Toolkit, device plugins, Enroot, and Pyxis. The line identifies the failing operation or subsystem, while the surrounding evidence decides whether it is the initiating fault or a downstream symptom.

The NVIDIA device plugin could not load libnvidia-ml.so, so it never registered nvidia.com/gpu with the kubelet. Every GPU pod on the node stays Pending with "Insufficient nvidia.com/gpu" while the node itself looks healthy. The failure becomes visible at this call site because the operation first requires the missing resource, valid state, healthy peer, or correct result. Earlier log lines and a known-good control carry more causal value than the final wrapper exception.

• The workload stops or loses forward progress after emitting "nvidia-device-plugin Failed to initialize NVML libnvidia-ml.so.1 cannot be found".
• A retry on the same configuration reproduces the failure because the causal state has not changed.
• The outer framework exception can hide the rank, node, allocation, or dependency that failed first.
• Increasing timeouts or reducing workload size can suppress the symptom without correcting the cause.

• NVIDIA Container Toolkit, device plugins, Enroot, and Pyxis
• production-shaped multi-accelerator workloads
• containerized and bare-metal deployments of the same stack

Apply the following checklist to a small reproduction: each box below is a positive signal that you are looking at this exact failure rather than a sibling in the same taxonomy.

• ✓Find the first occurrence of "nvidia-device-plugin Failed to initialize NVML libnvidia-ml.so.1 cannot be found" and preserve at least 100 lines before it.
• ✓Identify which rank, node, device, or process emitted the first related warning.
• ✓the library lives on the host and is mounted into the plugin container. A failure here means the driver is not installed on the host, the driver daemonset has not finished, or the volume mount / LD_LIBRARY_PATH does not reach it. Check driver readiness first, the plugin legitimately fails until the driver is up.
• ✓Repeat the same input after the targeted change and require the signature to disappear.
• ✓Resume from pod scheduling, once the resource is advertised only after the control passes.

nvidia-device-plugin Failed to initialize NVML libnvidia-ml.so.1 cannot be found

Timestamps and exact values vary across runs, but the pattern. An info-level start, an early WARN, an ERROR carrying the symptom. Is the actual fingerprint you should alert on. The Denpex platform flags this combination automatically.

check the plugin pod logs (kubectl logs -n gpu-operator -l app=nvidia-device-plugin-daemonset) and confirm the node advertises the resource: kubectl describe node <node> | grep nvidia.com/gpu. This changes the condition in the causal diagnosis instead of hiding the outer exception. The repeated control proves ownership before recovery from pod scheduling, once the resource is advertised.

# Preserve evidence before restarting
kubectl describe pod <pod>
kubectl get events --sort-by=.lastTimestamp
kubectl logs <pod> --all-containers

# Find the exact signature in the complete log
rg -n -F -- "nvidia-device-plugin Failed to initialize NVML libnvidia-ml.so.1 cannot be found" <log-file>

Adapt the snippet to your framework. The same pattern holds for PyTorch Lightning, Hugging Face Trainer, DeepSpeed, Megatron-LM, and vLLM training wrappers. Where the wrapper exposes a config flag (for examplelr_scheduler_type in Trainer), prefer the flag over the imperative API to keep the schedule declarative and reproducible.

Select the traceback or open the failed terminal, then run Denpex locally to see the initiating rank, collateral failures, exact fix, and verification command without uploading the log.

• The NVIDIA device plugin could not load libnvidia-ml.so, so it never registered nvidia.com/gpu with the kubelet. Every GPU pod on the node stays Pending with "Insufficient nvidia.com/gpu" while the node itself looks healthy.
• The decisive evidence is the first log line that precedes "nvidia-device-plugin Failed to initialize NVML libnvidia-ml.so.1 cannot be found" and differs from a healthy run.
• A nearby failure remains a competing hypothesis until a control separates configuration, capacity, transport, and hardware causes.