-
Notifications
You must be signed in to change notification settings - Fork 1.7k
Tensorboard pip package doesn't launch tensorboard on RHEL 7-ppc64le #748
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
Comments
Can you provide a little more detail on exactly what you're expecting to work? E.g. what did you do to install the freshly built pip package, and then how were you trying to execute the As far as I'm aware, the pip package building script isn't supposed to bundle the bazel-bin/tensorboard/tensorboard executable or the external dependencies into the wheel file (it did used to include some external sources prior to #591 but that was intentionally removed). The wheel file should pretty much just be the python source code plus compiled versions of our HTML/JS. When you
FWIW, the DMLC tensorboard is a derivative project that's not actually affiliated with us (yes, even though they're |
TensorBoard uses bleach, which depends on html5lib, which made a big breaking change at some point in the past. Since upgrading these dependencies is not possible, we'll shade them under the `tensorboard._vendor` namespace, which is magically created by our build_pip_package script, which was rewritten, to support smoke testing and dash. The new process for building the TensorBoard pip package is: ```sh rm -rf /tmp/tensorboard bazel run //tensorboard/pip_package:build_pip_package pip install -U /tmp/tensorboard/*py2*.pip ``` Fixes tensorflow/tensorflow#16424 Fixes tensorflow#427 Fixes tensorflow#588 Closes tensorflow#607 Could help tensorflow#748
TensorBoard uses bleach, which depends on html5lib, which made a big breaking change at some point in the past. Since upgrading these dependencies is not possible, we'll shade them under the `tensorboard._vendor` namespace, which is magically created by our build_pip_package script, which was rewritten, to support smoke testing and dash. The new process for building the TensorBoard pip package is: ```sh rm -rf /tmp/tensorboard bazel run //tensorboard/pip_package:build_pip_package pip install -U /tmp/tensorboard/*py2*.pip ``` Fixes tensorflow/tensorflow#16424 Fixes #427 Fixes #588 Closes #607 Could help #748
Closing as a support issue over 1 year old. Please re-open if this is still an issue. |
System:
Red Hat Enterprise Linux Server release 7.4 (Maipo), ppc64le
Tensorboard:
Built from source tag 0.4.0-rc3
Bazel:
0.5.4
Build Steps:
bazel build //tensorboard/pip_package:build_pip_package
bazel-bin/tensorboard/pip_package/build_pip_package /tmp
Issue:
The wheel file created when installed, doesn't copy the bazel-bin/tensorboard/tensorboard executable. And obviously, whatever is needed by this executable like tensorboard.runfiles is also not copied. Also the external pre-requisites like protobuf, html5lib, werkzeug, paper-, iron- are not copied into the wheel file.
On x86, I tried installing two packages tensorboard and tensorflow-tensorboard using pip. And 1st one provides complete package including all the dependencies. It looks like it comes from https://github.com/dmlc/tensorboard while the 2nd one is similar to what I see on ppc64le which came from https://github.com/tensorflow/tensorboard.
So, not sure if this is the bug in tensorflow-tensorboard package or the package is still evolving.
Meanwhile I tried modifying build_pip_package.sh to copy all the external dependencies and it worked.
The text was updated successfully, but these errors were encountered: