@ChrisRahme i think it’s the OS this time. I’m trying to install this within docker which has alpine as base image
Yup that’s it then, as mentioned above:
There is no Alpine here
@lis why you are using alpine image? are you trying to create the python environment in docker? or in what context you are using alpine? and in our previous conversation you have’t mention that you are trying to install rasa open source using docker, then my suggestion and solution was different.
@nik202 extremely sorry for not providing the full details. Yes, I was trying to create the python environment in docker with a python base image built over alpine OS.
1 Like
No worries, this detail isn’t that important if you want a local installation of Rasa. You just can’t install Rasa on Alpine, Docker or not (even the Rasa docs mention it is only possible on Windows, macOs, and Ubuntu).
But if you want to install Rasa on a server to be connected to a website or app shared with many users over the Internet, you will have to look at a server installation of Rasa like the Docker-Compose method.
Right @ChrisRahme, I’ve changed the base image to ubuntu. I’m trying something similar to the docker-compose method. With the docker-compose method, I had to automate the process of building an action server image using github actions/jenkins etc. What I am trying to achieve instead is to have Dockerfiles that would build container for rasa server and rasa action server (one in the main project directory where all the .yml files are present and another inside the actions folder). Write commands in a Makefile to build these containers. To achieve this I’m using a ubuntu-python as base image (ubuntu v18.04, python v3.8) I’m running pip install to install rasa==2.8.13 in one Dockerfile and pip install rasa-sdk==2.8.1 in the Dockerfile that’s present inside the actions folder. Idk if this is the right approach, but Makefile is the entrypoint of the platform I’m using to deploy this rasa app (Basically I was not able to do the github actions/jenkins setup with that platform)
I’m installing rasa and rasa-sdk as pointed out before.
Rasa Version : 2.8.13
Minimum Compatible Version: 2.8.9
Rasa SDK Version : 2.8.1
Rasa X Version : None
Python Version : 3.8.0
Operating System : Linux-5.10.60.1-microsoft-standard-WSL2-x86_64-with-glibc2.27
Python Path : /usr/src/python-app/venv/bin/python3
and I’m getting this error:
tensorflow.python.framework.errors_impl.AlreadyExistsError: Another metric with the same name already exists.
@lis For me every information is important and thanks for clearing the doubts, one last and simple question to you are trying to install rasa open-source on a local machine using docker? If yes what installation step you are using ? If not then what you trying to archived using docker.
@lis For you infromation, I had installed rasa open source, with rasa action server using docker (local machine) and also rasa X using docker and complete CI/CD pipeline using Githubactions(on server Ubuntu).
And right now me totally confused what you trying to archieve. Thanks.
1 Like
No idea what’s happening here. Can you share the Dockerfile?
I also found this: python - Error importing tensorflow "AlreadyExistsError: Another metric with the same name already exists." - Stack Overflow
Hi @nik202 , I’m trying to install rasa and rasa-sdk on a sever using Docker. Basically, my rasa application should be shipped to the cloud. To push an application to a cloud, we are using a platform that is internal. The platform is based out of CaaS. So this is a CaaS deployment to the cloud. This platform reads the dockerfile, executes the commands in it and creates a final build image that will be ported to the cloud to run the final container. So the problem now is, it’s throwing an error while trying to ship it to the cloud. So, I’m trying to replicate the error in my local to see what can be done to resolve this. When I started out, it was the base image itself (because it was a alpine OS with python 3.9). Now I’m using another base image (ubuntu with python 3.8) It’s trying to build from this base image (FROM docker--release..corp.**.com//python_v3:9.2). The files and folders are copied to this base container, followed by running a pip install inside it. This is when I’m getting this error. I tried checking what’s happening during installation - it installs tensorflow v2.6.2 first and then installs tensorflow v2.6.1.
Hi @ChrisRahme, I tried out the solution suggested in that stackoverflow already. It doesn’t seem to work for me.
Please share the Dockerfile
@ChrisRahme that issue was solved with --no-deps while running pip install. Thanks for your support., Anyways, this is my dockerfile -
FROM docker-****-release.****.corp.******.com/****/python_v3:9.2
COPY actions/ ${PYTHON_APP_FOLDER}/actions
COPY data/ ${PYTHON_APP_FOLDER}/data
COPY models/ ${PYTHON_APP_FOLDER}/models
COPY secrets/ ${PYTHON_APP_FOLDER}/secrets
COPY utils/ ${PYTHON_APP_FOLDER}/utils
COPY config.yml ${PYTHON_APP_FOLDER}
COPY credentials.yml ${PYTHON_APP_FOLDER}
COPY domain.yml ${PYTHON_APP_FOLDER}
COPY endpoints.yml ${PYTHON_APP_FOLDER}
COPY run_servers.sh ${PYTHON_APP_FOLDER}
COPY pip.conf ${PYTHON_APP_FOLDER}
COPY requirements.txt ${PYTHON_APP_FOLDER}
WORKDIR ${PYTHON_APP_FOLDER}
RUN ["pip3", "install", "--upgrade", "pip"]
RUN ["pip3", "install", "--no-deps", "-r", "requirements.txt"]
CMD ["sh", "run_servers.sh"]
1 Like
Thanks for sharing the solution 