vllm collect-env command collects detailed information about your environment, which is useful for debugging and reporting issues.
Usage
What it collects
The command gathers information about:- vLLM version - Installed vLLM version and commit hash
- Python version - Python interpreter version
- PyTorch version - PyTorch and CUDA versions
- GPU information - NVIDIA/AMD GPU model, driver version, CUDA/ROCm version
- CPU information - CPU model and architecture
- Operating system - OS type and version
- Environment variables - Relevant vLLM environment variables
- Dependencies - Key dependency versions (transformers, xformers, flash-attn, etc.)
Example output
When to use
Usevllm collect-env when:
Reporting issues
Include this output when filing GitHub issues
Debugging
Verify your environment setup and dependencies
Version checking
Quickly check installed versions of vLLM and dependencies
Support requests
Provide environment details when asking for help
Common issues detected
The command can help identify:- CUDA/driver mismatch - Incompatible CUDA and driver versions
- Missing dependencies - Required packages not installed
- Wrong Python version - Python version outside supported range (3.10-3.13)
- GPU not detected - GPU hardware or driver issues
Saving output to file
Related
- Development setup - Setting up a development environment
- Installation - Installation instructions for different platforms
- GitHub Issues - Report bugs and issues