> [build_triton 1/1] RUN --mount=type=cache,target=/root/.cache/ccache if [ "1" = "1" ]; then mkdir -p libs && cd libs && python3 -m p ip install ninja cmake wheel pybind11 && git clone https://github.com/OpenAI/triton.git && cd triton && git checkout "e192dba" && cd py thon && python3 setup.py bdist_wheel --dist-dir=/install; else mkdir -p /install; fi:
5.519 Collecting ninja
10.54 Downloading ninja-1.11.1.1-py2.py3-none-manylinux1_x86_64.manylinux_2_5_x86_64.whl.metadata (5.3 kB)
10.64 Collecting cmake
10.65 Downloading cmake-3.30.4-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (6.4 kB)
10.65 Requirement already satisfied: wheel in /opt/conda/envs/py_3.9/lib/python3.9/site-packages (0.43.0)
10.67 Collecting pybind11
10.67 Downloading pybind11-2.13.6-py3-none-any.whl.metadata (9.5 kB)
10.68 Downloading ninja-1.11.1.1-py2.py3-none-manylinux1_x86_64.manylinux_2_5_x86_64.whl (307 kB)
10.71 Downloading cmake-3.30.4-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (26.9 MB)
11.54 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 26.9/26.9 MB 32.5 MB/s eta 0:00:00
11.54 Downloading pybind11-2.13.6-py3-none-any.whl (243 kB)
11.77 Installing collected packages: ninja, pybind11, cmake
12.45 Successfully installed cmake-3.30.4 ninja-1.11.1.1 pybind11-2.13.6
12.45 WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager, possibly rend ering your system unusable.It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv. Use the --root-user-action option if you know what you are doing and want to suppress this warning.
12.50 Cloning into 'triton'...
41.57 Note: switching to 'e192dba'.
41.57
41.57 You are in 'detached HEAD' state. You can look around, make experimental
41.57 changes and commit them, and you can discard any commits you make in this
41.57 state without impacting any branches by switching back to a branch.
41.57
41.57 If you want to create a new branch to retain commits you create, you may
41.57 do so (now or later) by using -c with the switch command. Example:
41.57
41.57 git switch -c <new-branch-name>
41.57
41.57 Or undo this operation with:
41.57
41.57 git switch -
41.57
41.57 Turn off this advice by setting config variable advice.detachedHead to false
41.57
41.57 HEAD is now at e192dba22 [AMD] Hoist Q out of the loop for FA optimization (#4666)
226.8 downloading and extracting https://anaconda.org/nvidia/cuda-nvcc/12.4.99/download/linux-64/cuda-nvcc-12.4.99-0.tar.bz2 ...
226.8 Traceback (most recent call last):
226.8 File "/vllm-workspace/libs/triton/python/setup.py", line 489, in <module>
226.8 download_and_copy(
226.8 File "/vllm-workspace/libs/triton/python/setup.py", line 288, in download_and_copy
226.8 file = tarfile.open(fileobj=open_url(url), mode="r|*")
226.8 File "/vllm-workspace/libs/triton/python/setup.py", line 209, in open_url
226.8 return urllib.request.urlopen(request, timeout=300)
226.8 File "/opt/conda/envs/py_3.9/lib/python3.9/urllib/request.py", line 214, in urlopen
226.8 return opener.open(url, data, timeout)
226.8 File "/opt/conda/envs/py_3.9/lib/python3.9/urllib/request.py", line 523, in open
226.8 response = meth(req, response)
226.8 File "/opt/conda/envs/py_3.9/lib/python3.9/urllib/request.py", line 632, in http_response
226.8 response = self.parent.error(
226.8 File "/opt/conda/envs/py_3.9/lib/python3.9/urllib/request.py", line 561, in error
226.8 return self._call_chain(*args)
226.8 File "/opt/conda/envs/py_3.9/lib/python3.9/urllib/request.py", line 494, in _call_chain
226.8 result = func(*args)
226.8 File "/opt/conda/envs/py_3.9/lib/python3.9/urllib/request.py", line 641, in http_error_default
226.8 raise HTTPError(req.full_url, code, msg, hdrs, fp)
226.8 urllib.error.HTTPError: HTTP Error 524:
------
1 warning found (use docker --debug to expand):
- UndefinedVar: Usage of undefined variable '$CPLUS_INCLUDE_PATH' (line 63)
Dockerfile.rocm:101
--------------------
100 | # Build triton wheel if `BUILD_TRITON = 1`
101 | >>> RUN --mount=type=cache,target=${CCACHE_DIR} \
102 | >>> if [ "$BUILD_TRITON" = "1" ]; then \
103 | >>> mkdir -p libs \
104 | >>> && cd libs \
105 | >>> && python3 -m pip install ninja cmake wheel pybind11 \
106 | >>> && git clone https://github.com/OpenAI/triton.git \
107 | >>> && cd triton \
108 | >>> && git checkout "${TRITON_BRANCH}" \
109 | >>> && cd python \
110 | >>> && python3 setup.py bdist_wheel --dist-dir=/install; \
111 | >>> # Create an empty directory otherwise as later build stages expect one
112 | >>> else mkdir -p /install; \
113 | >>> fi
114 |
--------------------
ERROR: failed to solve: process "/bin/sh -c if [ \"$BUILD_TRITON\" = \"1\" ]; then mkdir -p libs && cd libs && python3 -m pip install ninja cmake wheel pybind11 && git clone https://github.com/OpenAI/triton.git && cd triton && git checkout \"${TRITON_BRANCH}\" && cd python && python3 setup.py bdist_wheel --dist-dir=/install; else mkdir -p /install; fi" did not complete successfully: exit code: 1
Vllm couldn't build successfully, I tried downloading https://anaconda.org/nvidia/cuda-nvcc/12.4.99/download/linux-64/cuda-nvcc-12.4.99-0.tar.bz2 and was able to do it. Haven't been able to find a fix yet, anyone else faced this issue too?
Before submitting a new issue...
[X] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.
Your current environment
How you are installing vllm
Vllm couldn't build successfully, I tried downloading
https://anaconda.org/nvidia/cuda-nvcc/12.4.99/download/linux-64/cuda-nvcc-12.4.99-0.tar.bz2
and was able to do it. Haven't been able to find a fix yet, anyone else faced this issue too?Before submitting a new issue...