Failed to load /opt/conda/lib/python3.11/site-packages/torchao/_C_cutlass_90a.abi3.so: Could not load this library: /opt/conda/lib/python3.11/site-packages/torchao/_C_cutlass_90a.abi3.so
Failed to load /opt/conda/lib/python3.11/site-packages/torchao/_C_mxfp8.cpython-310-x86_64-linux-gnu.so: Could not load this library: /opt/conda/lib/python3.11/site-packages/torchao/_C_mxfp8.cpython-310-x86_64-linux-gnu.so
W0725 03:04:42.333000 1 site-packages/torch/utils/_pytree.py:630] <enum 'KernelPreference'> is an Enum subclass and is now natively supported by torch.compile as an opaque value type. Calling register_constant() on Enum subclasses is deprecated and will be an error in a future release.
W0725 03:04:43.004000 1 site-packages/torch/utils/_pytree.py:630] <enum 'ScaleCalculationMode'> is an Enum subclass and is now natively supported by torch.compile as an opaque value type. Calling register_constant() on Enum subclasses is deprecated and will be an error in a future release.
{"@timestamp": "2026-07-25T03:04:55.950Z", "level": "INFO", "module": "xinference.core.worker", "pid": 1, "role": "worker", "address": "192.168.1.77:29395", "node": "njue-poweredge-r730", "message": "Starting metrics export server at 192.168.1.77:31001"}
{"@timestamp": "2026-07-25T03:04:55.951Z", "level": "INFO", "module": "xinference.core.worker", "pid": 1, "role": "worker", "address": "192.168.1.77:29395", "node": "njue-poweredge-r730", "message": "Checking metrics export server..."}
{"@timestamp": "2026-07-25T03:04:57.006Z", "level": "INFO", "module": "xinference.core.worker", "pid": 1, "role": "worker", "address": "192.168.1.77:29395", "node": "njue-poweredge-r730", "message": "Metrics server is started at: http://192.168.1.77:31001"}
{"@timestamp": "2026-07-25T03:04:57.006Z", "level": "INFO", "module": "xinference.core.worker", "pid": 1, "role": "worker", "address": "192.168.1.77:29395", "node": "njue-poweredge-r730", "message": "Purge cache directory: /data/cache"}
{"@timestamp": "2026-07-25T03:04:57.336Z", "level": "INFO", "module": "xinference.core.worker", "pid": 1, "role": "worker", "address": "192.168.1.77:29395", "node": "njue-poweredge-r730", "message": "Connected to supervisor as a fresh worker"}
{"@timestamp": "2026-07-25T03:04:57.542Z", "level": "INFO", "module": "xinference.core.worker", "pid": 1, "role": "worker", "address": "192.168.1.77:29395", "node": "njue-poweredge-r730", "message": "Xinference worker 192.168.1.77:29395 started"}
{"@timestamp": "2026-07-25T03:04:57.543Z", "level": "WARNING", "module": "xinference.core.worker", "pid": 1, "role": "worker", "address": "192.168.1.77:29395", "node": "njue-poweredge-r730", "message": "Startup GPU orphan cleanup skipped: pynvml init failed or timed out (10s). If the previous worker left vLLM orphans, launch may hang. Manual check: nvidia-smi + ps -eo pid,ppid,cmd | grep vllm"}
