Sudden TypeError - Descriptors cannot not be created directly

[rialco]

Summary

I was using the image python:3.9.13-slim to deploy a container that uses tensorflow and other libraries. Everything was okay but with the latest image release at the moment of writting: Digest:sha256:5e652db0ae12b5e5e538466f5a6ca4ae0b4e5f0c647cbfbd1071ac75edb3785d I began to have issues with the container even though I didn't change a thing.

I was prompt with the following error when the container was starting:

Traceback (most recent call last):
neural-network           |   File "/app/app.py", line 6, in <module>
neural-network           |     from model import Model
neural-network           |   File "/app/model.py", line 3, in <module>
neural-network           |     from tensorflow import keras
neural-network           |   File "/opt/venv/lib/python3.9/site-packages/tensorflow/__init__.py", line 37, in <module>
neural-network           |     from tensorflow.python.tools import module_util as _module_util
neural-network           |   File "/opt/venv/lib/python3.9/site-packages/tensorflow/python/__init__.py", line 37, in <module>
neural-network           |     from tensorflow.python.eager import context
neural-network           |   File "/opt/venv/lib/python3.9/site-packages/tensorflow/python/eager/context.py", line 29, in <module>
neural-network           |     from tensorflow.core.framework import function_pb2
neural-network           |   File "/opt/venv/lib/python3.9/site-packages/tensorflow/core/framework/function_pb2.py", line 16, in <module>
neural-network           |     from tensorflow.core.framework import attr_value_pb2 as tensorflow_dot_core_dot_framework_dot_attr__value__pb2
neural-network           |   File "/opt/venv/lib/python3.9/site-packages/tensorflow/core/framework/attr_value_pb2.py", line 16, in <module>
neural-network           |     from tensorflow.core.framework import tensor_pb2 as tensorflow_dot_core_dot_framework_dot_tensor__pb2
neural-network           |   File "/opt/venv/lib/python3.9/site-packages/tensorflow/core/framework/tensor_pb2.py", line 16, in <module>
neural-network           |     from tensorflow.core.framework import resource_handle_pb2 as tensorflow_dot_core_dot_framework_dot_resource__handle__pb2
neural-network           |   File "/opt/venv/lib/python3.9/site-packages/tensorflow/core/framework/resource_handle_pb2.py", line 16, in <module>
neural-network           |     from tensorflow.core.framework import tensor_shape_pb2 as tensorflow_dot_core_dot_framework_dot_tensor__shape__pb2
neural-network           |   File "/opt/venv/lib/python3.9/site-packages/tensorflow/core/framework/tensor_shape_pb2.py", line 36, in <module>
neural-network           |     _descriptor.FieldDescriptor(
neural-network           |   File "/opt/venv/lib/python3.9/site-packages/google/protobuf/descriptor.py", line 560, in __new__
neural-network           |     _message.Message._CheckCalledFromGeneratedFile()
neural-network           | TypeError: Descriptors cannot not be created directly.
neural-network           | If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
neural-network           | If you cannot immediately regenerate your protos, some other possible workarounds are:
neural-network           |  1. Downgrade the protobuf package to 3.20.x or lower.
neural-network           |  2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).
neural-network           | 
neural-network           | More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates

Environment

I am running docker desktop 4.8.2 in a Mac M1 machine.

My dockerfile looks like this:

FROM python:3.9.13-slim

WORKDIR /app

ENV VIRTUAL_ENV=/opt/venv
RUN python -m venv $VIRTUAL_ENV
ENV PATH="$VIRTUAL_ENV/bin:$PATH"

RUN apt-get -y update \
    && apt install -y build-essential libpq-dev

RUN pip install --upgrade pip
RUN pip install redis
RUN pip install nltk
RUN pip install numpy
RUN pip install pandas
RUN pip install sqlalchemy
# x86_64
# RUN pip install tensorflow
# RUN pip install psycopg2-binary
# aarch64
RUN pip install tensorflow-aarch64 -f https://tf.kmtea.eu/whl/stable.html
RUN pip install psycopg2

COPY . .

CMD ["python", "-u", "app.py"]

Current workaround

I found a solution in a not so popular stackoverflow comment in which the person suggested to:

pip uninstall protobuf
pip install --no-binary protobuf protobuf

[wglambert]

The workaround you've got seems fine then for the error

neural-network           | If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
neural-network           | If you cannot immediately regenerate your protos, some other possible workarounds are:
neural-network           |  1. Downgrade the protobuf package to 3.20.x or lower.
neural-network           |  2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).
neural-network           | 
neural-network           | More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates

Going to close since the error is related to the updated Python version and isn't an issue with the image itself
If you need further help you could also try asking over at the Docker Community Forums, Docker Community Slack, or Stack Overflow. Since these repos aren't really a user-help forum