I m using quick install/helm for rasa server’s deployment, and everything was working properly including model upload on server rasa x, until i upgraded to 2.8.2 (bug, not starting) and downgraded back to 2.7.0. After the deployment, since i m using local GIT and not github, i can’t train the model on server rasa x and i need to upload the model manually using Add model → Upload files…before i could upload the models without any problems, but now it’s failing…i tried to downgrade to 2.6.0, restart docker, delete namespace, images…without success.
The upload via API is not working.
I suspect the upgrade to 2.8.0 has messed everything up. Please advise if there is any alternative method to upload the model to docker pod, or how to solve this problem.
i was able to upload the models all time, then suddenly stopped working, i don’t know what’s wrong as there is no clue about the source of this problem.
Today, as alternative solution, i trained/activated the model using github, but when i type anything the bot doesn’t answer (keeps typing always, all pods are running, no errors…). I m sure there is something occurred in Rasa server, but i don’t know what it is or how to fix it. Note that everything was working properly then this problem started.
it seems quick install was installing the latest versions of Rasa (2.8.0 / 0.42) which have bug (not working also on local mode). I solved the problem by forcing the usage of 2.7.0 and 0.41.2 (RASA_X_VERSION and RASA_VERSION variables).