-
Notifications
You must be signed in to change notification settings - Fork 166
Enable AWSSDK on Linux by statically linking OpenSSL and cURL #421
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
ef8968f
to
0453f12
Compare
ae88807
to
21abff4
Compare
@ejguan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM! Thanks for making all these changes!
Can you elaborate on "Make our testing CI running with AWSSDK enabled by default to save more time/money to run CI tests"? How does that work?
@@ -65,7 +62,7 @@ jobs: | |||
run: | | |||
python setup.py install | |||
env: | |||
BUILD_S3: ${{ matrix.with-s3 }} | |||
BUILD_S3: 1 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is there any situation where we do not enable BUILD_S3
?
It looks like only when the setup is Windows with conda, why is that?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
For binary releases, it's just because there are too many Error that Idk how to solve with AWS enabled for conda on windows.
For CI testing, you can treat it as wheels, we could enable testing with awssdk.
Generally, the reason I think it makes sense to always enable AWSSDK in our testing CI is the >95% tests are the same with or without AWSSDK enabled.
By simply removing one test case of |
…h#421) Summary: This PR adds all supports to enable awssdk for linux releases - The linux release workflow starts to use `pytorch/manylinux-cpu` docker to align with compiler version with PyTorch Core - Add shell script to install OpenSSL and cURL on docker image (This step can be improved if TorchData wants to own a new docker image with these two libraries pre-built) - Update `torchdata/csrc/CMakeLists.txt` to respect the static OpenSSL and cURL - Correct `thrid_party/CMakeLists.txt` without adding redundant dependencies for `_torchdata` - Add `auditwheel show` to validate the binaries are `manylinux_2_17` (alias of `manylinux2014`) - Update Readme for release - Make our testing CI running with AWSSDK enabled by default to save more time/money to run CI tests - Remove redundant `numpy` dependency introduced by `tf_record`. And, fix tests in this file See the workflow for night release: https://github.com/pytorch/data/actions/runs/2391843245 See the `manylinux` version: https://github.com/pytorch/data/runs/6612861049?check_suite_focus=true#step:9:67 `manylinux_2_17_x86_64` is the alias of `manylinux_2014` Pull Request resolved: pytorch#421 Reviewed By: NivekT Differential Revision: D36641092 Pulled By: ejguan fbshipit-source-id: 349bfd896ee0db01eea849580984f4000ca2bc3f
Summary: This PR adds all supports to enable awssdk for linux releases - The linux release workflow starts to use `pytorch/manylinux-cpu` docker to align with compiler version with PyTorch Core - Add shell script to install OpenSSL and cURL on docker image (This step can be improved if TorchData wants to own a new docker image with these two libraries pre-built) - Update `torchdata/csrc/CMakeLists.txt` to respect the static OpenSSL and cURL - Correct `thrid_party/CMakeLists.txt` without adding redundant dependencies for `_torchdata` - Add `auditwheel show` to validate the binaries are `manylinux_2_17` (alias of `manylinux2014`) - Update Readme for release - Make our testing CI running with AWSSDK enabled by default to save more time/money to run CI tests - Remove redundant `numpy` dependency introduced by `tf_record`. And, fix tests in this file See the workflow for night release: https://github.com/pytorch/data/actions/runs/2391843245 See the `manylinux` version: https://github.com/pytorch/data/runs/6612861049?check_suite_focus=true#step:9:67 `manylinux_2_17_x86_64` is the alias of `manylinux_2014` Pull Request resolved: #421 Reviewed By: NivekT Differential Revision: D36641092 Pulled By: ejguan fbshipit-source-id: 349bfd896ee0db01eea849580984f4000ca2bc3f
…h#421) Summary: This PR adds all supports to enable awssdk for linux releases - The linux release workflow starts to use `pytorch/manylinux-cpu` docker to align with compiler version with PyTorch Core - Add shell script to install OpenSSL and cURL on docker image (This step can be improved if TorchData wants to own a new docker image with these two libraries pre-built) - Update `torchdata/csrc/CMakeLists.txt` to respect the static OpenSSL and cURL - Correct `thrid_party/CMakeLists.txt` without adding redundant dependencies for `_torchdata` - Add `auditwheel show` to validate the binaries are `manylinux_2_17` (alias of `manylinux2014`) - Update Readme for release - Make our testing CI running with AWSSDK enabled by default to save more time/money to run CI tests - Remove redundant `numpy` dependency introduced by `tf_record`. And, fix tests in this file See the workflow for night release: https://github.com/pytorch/data/actions/runs/2391843245 See the `manylinux` version: https://github.com/pytorch/data/runs/6612861049?check_suite_focus=true#step:9:67 `manylinux_2_17_x86_64` is the alias of `manylinux_2014` Pull Request resolved: pytorch#421 Reviewed By: NivekT Differential Revision: D36641092 Pulled By: ejguan fbshipit-source-id: 349bfd896ee0db01eea849580984f4000ca2bc3f
This PR adds all supports to enable awssdk for linux releases
pytorch/manylinux-cpu
docker to align with compiler version with PyTorch Coretorchdata/csrc/CMakeLists.txt
to respect the static OpenSSL and cURLthrid_party/CMakeLists.txt
without adding redundant dependencies for_torchdata
auditwheel show
to validate the binaries aremanylinux_2_17
(alias ofmanylinux2014
)See the successful workflow for night release: https://github.com/pytorch/data/actions/runs/2391843245
See the
manylinux
version: https://github.com/pytorch/data/runs/6612861049?check_suite_focus=true#step:9:67manylinux_2_17_x86_64
is the alias ofmanylinux_2014