RunPod•4mo ago

worker-vllm build fails

I am getting the following error when building the new worker-vllm image with my model.

 => ERROR [vllm-base 6/7] RUN --mount=type=secret,id=HF_TOKEN,required=false     if [ -f /run/secrets/HF_TOKEN ]; then         export HF_TOKEN=$(cat /run/secrets/HF_TOKEN);     fi &&     if [ -n "Pate  10.5s
------
 > [vllm-base 6/7] RUN --mount=type=secret,id=HF_TOKEN,required=false     if [ -f /run/secrets/HF_TOKEN ]; then         export HF_TOKEN=$(cat /run/secrets/HF_TOKEN);     fi &&     if [ -n "PatentPilotAI/mistral-7b-patent-instruct-v2" ]; then         python3 /download_model.py;     fi:
#10 9.713 Traceback (most recent call last):
#10 9.713   File "/download_model.py", line 4, in <module>
#10 9.715     from vllm.model_executor.weight_utils import prepare_hf_model_weights, Disabledtqdm
#10 9.715   File "/vllm-installation/vllm/model_executor/__init__.py", line 2, in <module>
#10 9.715     from vllm.model_executor.model_loader import get_model
#10 9.715   File "/vllm-installation/vllm/model_executor/model_loader.py", line 10, in <module>
#10 9.715     from vllm.model_executor.weight_utils import (get_quant_config,
#10 9.715   File "/vllm-installation/vllm/model_executor/weight_utils.py", line 18, in <module>
#10 9.715     from vllm.model_executor.layers.quantization import (get_quantization_config,
#10 9.715   File "/vllm-installation/vllm/model_executor/layers/quantization/__init__.py", line 4, in <module>
#10 9.716     from vllm.model_executor.layers.quantization.awq import AWQConfig
#10 9.716   File "/vllm-installation/vllm/model_executor/layers/quantization/awq.py", line 6, in <module>
#10 9.716     from vllm._C import ops
#10 9.716 ImportError: libcuda.so.1: cannot open shared object file: No such file or directory

 => ERROR [vllm-base 6/7] RUN --mount=type=secret,id=HF_TOKEN,required=false     if [ -f /run/secrets/HF_TOKEN ]; then         export HF_TOKEN=$(cat /run/secrets/HF_TOKEN);     fi &&     if [ -n "Pate  10.5s
------
 > [vllm-base 6/7] RUN --mount=type=secret,id=HF_TOKEN,required=false     if [ -f /run/secrets/HF_TOKEN ]; then         export HF_TOKEN=$(cat /run/secrets/HF_TOKEN);     fi &&     if [ -n "PatentPilotAI/mistral-7b-patent-instruct-v2" ]; then         python3 /download_model.py;     fi:
#10 9.713 Traceback (most recent call last):
#10 9.713   File "/download_model.py", line 4, in <module>
#10 9.715     from vllm.model_executor.weight_utils import prepare_hf_model_weights, Disabledtqdm
#10 9.715   File "/vllm-installation/vllm/model_executor/__init__.py", line 2, in <module>
#10 9.715     from vllm.model_executor.model_loader import get_model
#10 9.715   File "/vllm-installation/vllm/model_executor/model_loader.py", line 10, in <module>
#10 9.715     from vllm.model_executor.weight_utils import (get_quant_config,
#10 9.715   File "/vllm-installation/vllm/model_executor/weight_utils.py", line 18, in <module>
#10 9.715     from vllm.model_executor.layers.quantization import (get_quantization_config,
#10 9.715   File "/vllm-installation/vllm/model_executor/layers/quantization/__init__.py", line 4, in <module>
#10 9.716     from vllm.model_executor.layers.quantization.awq import AWQConfig
#10 9.716   File "/vllm-installation/vllm/model_executor/layers/quantization/awq.py", line 6, in <module>
#10 9.716     from vllm._C import ops
#10 9.716 ImportError: libcuda.so.1: cannot open shared object file: No such file or directory

18 Replies

Casper.•4mo ago

@Alpay Ariyak I'm not sure what the exact issue is here. Do you need a GPU to build the image? I checked out commit 2b5b8dfb61e32d221bc8ce49f98ec74698154a6e to get it working for now. Seems latest release is broken somehow

ashleyk•4mo ago

Its a regression, was fixed previously.

Casper.•4mo ago

Actually, I had to go even further back to get it working 😅 Just pushing my new serverless image now. Would love for this to be fixed so that I can upgrade bfeb60c54eaad2eeffa9741ce7600eb30e573698

Alpay Ariyak•4mo ago

Checking this now This has to do with vLLM's updates Seems to only be affecting AWQ and potentially other quants

Alpay Ariyak•4mo ago

So this is why https://github.com/vllm-project/vllm/blob/929b4f2973ec6a53ea4f0f03d21147ef8b8278be/vllm/model_executor/weight_utils.py#L85-L122 Love the # TODO(woosuk): Move this to other place. lol The function is not used in that file in any way

GitHub

vllm/vllm/model_executor/weight_utils.py at 929b4f2973ec6a53ea4f0f0...

A high-throughput and memory-efficient inference and serving engine for LLMs - vllm-project/vllm

Alpay Ariyak•4mo ago

Hey @Casper. could you try with image 0.3.1 as base In this line instead of 0.3.0 https://github.com/runpod-workers/worker-vllm/blob/717343b0ad4d8a4ea76626c52b473619c646e30b/Dockerfile#L2 Pushed a new image, should work

Casper.•4mo ago

My model is not quantized I’ll try it later

Alpay Ariyak•4mo ago

Got you Should fix either way

Casper.•4mo ago

Getting same error ImportError: libcuda.so.1: cannot open shared object file: No such file or directory

Alpay Ariyak•4mo ago

What’s the full error

Casper.•4mo ago

Same as I pasted above

 > [vllm-base 6/7] RUN --mount=type=secret,id=HF_TOKEN,required=false     if [ -f /run/secrets/HF_TOKEN ]; then         export HF_TOKEN=$(cat /run/secrets/HF_TOKEN);     fi &&     if [ -n "PatentPilotAI/mistral-7b-patent-instruct-v2" ]; then         python3 /download_model.py;     fi:
#9 10.13 Traceback (most recent call last):
#9 10.13   File "/download_model.py", line 4, in <module>
#9 10.13     from vllm.model_executor.weight_utils import prepare_hf_model_weights, Disabledtqdm
#9 10.13   File "/vllm-installation/vllm/model_executor/__init__.py", line 2, in <module>
#9 10.13     from vllm.model_executor.model_loader import get_model
#9 10.13   File "/vllm-installation/vllm/model_executor/model_loader.py", line 10, in <module>
#9 10.13     from vllm.model_executor.weight_utils import (get_quant_config,
#9 10.13   File "/vllm-installation/vllm/model_executor/weight_utils.py", line 18, in <module>
#9 10.13     from vllm.model_executor.layers.quantization import QuantizationConfig
#9 10.13   File "/vllm-installation/vllm/model_executor/layers/quantization/__init__.py", line 4, in <module>
#9 10.13     from vllm.model_executor.layers.quantization.awq import AWQConfig
#9 10.13   File "/vllm-installation/vllm/model_executor/layers/quantization/awq.py", line 6, in <module>
#9 10.13     from vllm._C import ops
#9 10.14 ImportError: libcuda.so.1: cannot open shared object file: No such file or directory
------
executor failed running [/bin/sh -c if [ -f /run/secrets/HF_TOKEN ]; then         export HF_TOKEN=$(cat /run/secrets/HF_TOKEN);     fi &&     if [ -n "$MODEL_NAME" ]; then         python3 /download_model.py;     fi]: exit code: 1

 > [vllm-base 6/7] RUN --mount=type=secret,id=HF_TOKEN,required=false     if [ -f /run/secrets/HF_TOKEN ]; then         export HF_TOKEN=$(cat /run/secrets/HF_TOKEN);     fi &&     if [ -n "PatentPilotAI/mistral-7b-patent-instruct-v2" ]; then         python3 /download_model.py;     fi:
#9 10.13 Traceback (most recent call last):
#9 10.13   File "/download_model.py", line 4, in <module>
#9 10.13     from vllm.model_executor.weight_utils import prepare_hf_model_weights, Disabledtqdm
#9 10.13   File "/vllm-installation/vllm/model_executor/__init__.py", line 2, in <module>
#9 10.13     from vllm.model_executor.model_loader import get_model
#9 10.13   File "/vllm-installation/vllm/model_executor/model_loader.py", line 10, in <module>
#9 10.13     from vllm.model_executor.weight_utils import (get_quant_config,
#9 10.13   File "/vllm-installation/vllm/model_executor/weight_utils.py", line 18, in <module>
#9 10.13     from vllm.model_executor.layers.quantization import QuantizationConfig
#9 10.13   File "/vllm-installation/vllm/model_executor/layers/quantization/__init__.py", line 4, in <module>
#9 10.13     from vllm.model_executor.layers.quantization.awq import AWQConfig
#9 10.13   File "/vllm-installation/vllm/model_executor/layers/quantization/awq.py", line 6, in <module>
#9 10.13     from vllm._C import ops
#9 10.14 ImportError: libcuda.so.1: cannot open shared object file: No such file or directory
------
executor failed running [/bin/sh -c if [ -f /run/secrets/HF_TOKEN ]; then         export HF_TOKEN=$(cat /run/secrets/HF_TOKEN);     fi &&     if [ -n "$MODEL_NAME" ]; then         python3 /download_model.py;     fi]: exit code: 1

Alpay Ariyak•4mo ago

Are you sure you’re using 0.3.1?

Casper.•4mo ago

Yeah it loads the new 0.3.1 [vllm-base 1/7] FROM docker.io/runpod/worker-vllm:base-0.3.1-cuda11.8.0 I'm on the latest commit d91ccb866fc784b81a558f0da44041a020ba54e0

Alpay Ariyak•4mo ago

I see what's going on

Casper.•4mo ago

I am building on a Macbook M2 btw Just for reference

Alpay Ariyak•4mo ago

Pushing new base image Done, try now

Casper.•4mo ago

Rebuilding now, let's see Build worked @Alpay Ariyak, thanks for fixing it

Alpay Ariyak•4mo ago

No problem, thanks for pointing it out!

Gaming

Programming

worker-vllm build fails