rasa_stack_quickstart weatherbot rasa_core Dockerfile pulling older version of requirements.txt for rasa_core about rasa_stack_quickstart HOT 5 CLOSED

@HaythamAmin
Yes, this repo used a branch from rasa core(specifically 0.9.0a3). It had 1.4 tensorflow requirement. Are you running on windows? Can you try to run pip install tensorfow and see if it works outside of the container?

from rasa_stack_quickstart.

I'm running this in an ubuntu server. That did not work unfortunately! I even altered the dockerfile in rasa_core and added a step to install Tensorflow before requirements.txt. Tensorflow got installed but the container crashed at the step after.

Dockerfile after update:

#Uses Latest Rasa Core.
RUN git clone -b 0.9.0a3 https://github.com/RasaHQ/rasa_core.git
WORKDIR /app/rasa_core
RUN pip install tensorflow
RUN pip install -r requirements.txt
RUN pip install -e .

Run stops with the following error:

Step 9/15 : WORKDIR /app/rasa_core
 ---> Using cache
 ---> af0d0494a109
Step 10/15 : RUN pip install tensorflow
 ---> Running in 599b8d1f755a
Collecting tensorflow
  Could not find a version that satisfies the requirement tensorflow (from versions: )
No matching distribution found for tensorflow
ERROR: Service 'rasa-core' failed to build: The command '/bin/sh -c pip install tensorflow' returned a non-zero code: 1

Maybe you need to update the requirements.txt file for it to work.

from rasa_stack_quickstart.

Ok, I got it!

The issue is in rasa_stack_quickstart/weatherbot/rasa_core/Dockerfile LINE 1 FROM python:3. This will force the container to install the latest Python 3 i.e python 3.7, we can confirm that by adding a Line to that Docker file to check for Python version before we proceed with running RUN pip install -r requirements.txt.

The issue with this is that Tensorflow does not support pthon 3.7 yet, we can confirm that by checking Tensorflow website and read through the different tensorflow python packages list and python version support.

Solution: Alter Dockerfile to specify python version. In my case I chose 3.5, so that my Dockerfile looks like this:

FROM python:3.5

RUN apt-get update && apt-get install -y \
    software-properties-common
RUN apt-get update && apt-get install -y \
    curl \
    git \
    python3 \
    python3-pip

RUN mkdir /app
ADD /data /data
RUN chmod -R 777 /data

WORKDIR /app
#RUN pip install rasa_core

#Uses Latest Rasa Core.
RUN git clone -b 0.9.0a3 https://github.com/RasaHQ/rasa_core.git
WORKDIR /app/rasa_core

RUN python -m pip install https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.8.0-cp35-cp35m-linux_x86_64.whl
RUN pip install -r requirements.txt
RUN pip install -e .

WORKDIR /data

EXPOSE 5005
ENTRYPOINT ["./entrypoint.sh"]

I've also included tensorflow url for pip to insure right package gets installed.

from rasa_stack_quickstart.

@HaythamAmin did this change work for you?
Looks like there was a change to python3 image 3 days back. They must have updated it to 3.7. If you have these changes in your fork feel free to submit a PR.

from rasa_stack_quickstart.

On it! Thanks for all your hard work :-)

from rasa_stack_quickstart.