How to mount the model file which is added as part of docker image in kubernetes deployment ?
The file is present in the instance running RASA OSS in /model folder, but it is not picking it up.
I see below mounts but not the model mount:
Mounts:
/.config from config-dir (rw)
/app/credentials.yml from rasa-configuration (ro,path="credentials.yml")
/app/endpoints.yml from rasa-configuration (ro,path="endpoints.yml")
/var/run/secrets/kubernetes.io/serviceaccount from default-token-tbmqm (ro)
@nik202 , can you please help here.
I have a dockerfile, which contains the model, as it is being trained during the docker file creation itself.
When I am using RASA helm charts to deploy the dockerimage, its not loading the model.
dockerfile:
RUN rasa train --fixed-model-name latest
ENTRYPOINT ["rasa"]
CMD [ "run","-m","/app/models/latest.tar.gz","--enable-api","--cors","'","*","'","--debug" ]
Logs from K8s pod:
/opt/venv/lib/python3.8/site-packages/rasa/shared/utils/io.py:99: UserWarning: No valid configuration given to load agent. Agent loaded with no model!
This is ONLY happening when, RasaX is enabled in the config.
# Rasa X / Enterprise settings
rasaX:
# -- Run Rasa X / Enterprise server
enabled: true
# -- Token Rasa X / Enterprise accepts as authentication token from other Rasa services
token: <token>
# -- Rasa X / Enterprise endpoint URL from which to pull the runtime config
useConfigEndpoint: false
# -- URL to Rasa X / Enterprise, e.g. http://rasa-x.mydomain.com:5002
url: <internal rasa-x url>
endpoints:
models:
enabled: false
token: <token>
useRasaXasModelServer:
enabled: false
tag: production
waitTimeBetweenPulls: 20
If I set enabled: false, the model is loading fine.