sudo apt update
sudo apt install docker.io docker-compose
sudo snap install microk8s --classic
sudo usermod -a -G microk8s $USER
sudo chown -f -R $USER ~/.kube
sudo microk8s enable dns storage helm3 registry dashboard ingress
cd $HOME/.kube
sudo microk8s config > config
alias kubectl='microk8s.kubectl'
alias helm='microk8s.helm3'
alias k="kubectl --namespace ask-edna"
alias h="helm --namespace ask-edna"
source ~/.bashrc
kubectl create namespace ask-edna
helm repo add rasa-x https://rasahq.github.io/rasa-x-helm
h install --values values.yml ask-edna rasa-x/rasa-x
The values.yml file is as below:
debugMode: true
rasax:
# initialUser is the user which is created upon the initial start of Rasa X
initialUser:
# password for the Rasa X user
password: "Password1"
# token Rasa X accepts as authentication token from other Rasa services
token: "Password1"
# jwtSecret which is used to sign the jwtTokens of the users
jwtSecret: "Password1"
# rasa: Settings common for all Rasa containers
rasa:
# token Rasa accepts as authentication token from other Rasa services
token: "Password1"
name: "xxxxxxxx/rasaxxxxxxxxx"
# tag refers to the Rasa image tag. If empty `.Values.rasa.version-full` is used.
tag: "1"
# RabbitMQ specific settings
rabbitmq:
# rabbitmq settings of the subchart
rabbitmq:
# password which is used for the authentication
password: "Password1"
# global settings of the used subcharts
global:
# postgresql: global settings of the postgresql subchart
postgresql:
# postgresqlPassword is the password which is used when the postgresqlUsername equals "postgres"
postgresqlPassword: "Password1"
# redis: global settings of the postgresql subchart
redis:
# password to use in case there no external secret was provided
password: "Password1"
app:
name: "xxxxxxxx/customactionserver"
tag: "1"
duckling:
enabled: true
# name of the Duckling image to use
name: "rasa/duckling"
# port on which duckling should run
port: 8000
# scheme by which duckling is accessible
scheme: http
nginx:
service:
type: LoadBalancer
port: 80
externalIPs: [100.xx.xx.xx]
If you notice we are using 2 custom images hosted in dockerhub:
for action server
rasa, as we are using custom components: Now for this we have used the below dockerfile:
FROM rasa/rasa:2.1.3-full
USER root
RUN pip install --trusted-host pypi.python.org fuzzywuzzy==0.18.0 rapidfuzz==0.13.4
We are able to access Rasax, upload and activate the model. But:
Github integration is failing with “Something went wrong”: deploy key added with write access in github
Found one possible cause: the custom rasa image file is not being created correctly… updated the docker file but still facing the same issue:
FROM rasa/rasa:2.1.3-full
USER root
COPY ./customComp /customComp/
ENV PYTHONPATH=$PYTHONPATH:/customComp/
RUN pip install --trusted-host pypi.python.org fuzzywuzzy==0.18.0 rapidfuzz==0.13.4
# Switch back to non-root to run code
USER 1001
@mloubser Thanks for taking some time out and replying.
To answer your question Nothing is wrong, the image is getting created correctly and pulled correctly at least all all the pods are up and running… my questions are:
Is the way in which I am creating the image and copying the customcomp correct?
What should i mention in the values.yml for the helm chart deployment in order to use this?
Because the problems that i have mentioned in the initial question still remains…
No response from uploaded model – which i think probably due to the custom comp issue
Cannot link github
Can’t even check the logs – not sure why this is happening
Screenshots to these are given in the original question.
This is something which I am already doing… can you please check the initial question of this post? It has the values.yml that i am using… I was worried if the custom component needs to specified in the values.yml
logs - screenshot is attached in the initial question of this thread, PFB the same
Can you use the docker image to train a model locally? To run a model locally? Any issues with the custom component won’t show up unless you try one of those.
No, the custom component doesn’t need to be specified in values.yml
Can you use the docker image to train a model locally? To run a model locally? Any issues with the custom component won’t show up unless you try one of those.
@dearc as i mentioned to you when we discussed about, your docker image is still incorrect.
you need to add an entrypoint because your docker image is not running as per the logs. Also using Docker UI is not going to provide you useful logs, use the docker command line instead to run your images. Using UI, requires you to provide an entrypoint to your image which is missing in the above Dockerfile you provided.
Rebuild the below image and use it to run. also how exactly are you running the docker image?
please try
docker run yourimagename rasa run
simply using the docker UI to run images will require you to add an entrypoint.
FROM rasa/rasa:2.1.3-full
USER root
COPY ./customComp /customComp/
ENV PYTHONPATH=$PYTHONPATH:/customComp/
RUN pip install --trusted-host pypi.python.org fuzzywuzzy==0.18.0 rapidfuzz==0.13.4
# Switch back to non-root to run code
USER 1001
ENTRYPOINT rasa run
