Python segmentation fault when using tensorflow-io==0.27.0 on MacOS

[abarclay06]

Issue:

Received Fatal Python error: Segmentation fault when running tensorboard using tensorflow-io version 0.27.0. This issue does not happen when using version 0.26.0. Was not able to reproduce the issue on a linux based docker image.

Specs:

This error was produced on my mac with the following specs, but it was also reproduced on a co-worker's mac.

ProductName: macOS ProductVersion: 11.6.5 BuildVersion: 20G527 kern.version: Darwin Kernel Version 20.6.0: Tue Feb 22 21:10:41 PST 2022; root:xnu-7195.141.26~1/RELEASE_X86_64

Python 3.9.6 (default, Apr 14 2022, 16:58:48) [Clang 12.0.5 (clang-1205.0.22.11)] on darwin

Minimum Reproducible Example:

Ran the following and received Python segmentation fault:

python -m venv ./.venv
source .venv/bin/activate
pip install tensorflow==2.10.0 tensorboard==2.10.0 tensorflow-io==0.27.0
tensorboard --logdir s3://ryft-public-sample-data/

Ran the above using tensorflow-io=0.26.0 and did not receive error.

[matemijolovic]

I'm having a very similar issue with Tensorflow IO on my machine.

Machine: Macbook Pro 2019 (i7) OS: MacOS Ventura 13.1 Python 3.9.13 or 3.8.15 Tensorflow 2.10.1 Tensorflow IO 0.27.0

When I try to run any code that uses Tensorflow IO (e.g. train run of a model that fetches inputs from S3) I get SIGSEGV. The exact same code works on Linux machines and in a Linux Docker container. If I downgrade to Tensorflow IO 0.26.0 everything works, although the official compatibility table says it shouldn't be compatible.

[jeongukjae]

https://github.com/tensorflow/io/issues/1828 Maybe related with this issue