Could not load model from S3 due to 'ContentLength'

Rasa Open Source
1

Hi,

I’m trying to fetch rasa trained model from S3 bucket in docker. I have provided the aws credentials in dockerfile. On running the below command:

rasa run -m 20220605-234203.tar.gz --remote-storage aws --enable-api -p 5005 --log-file rasa-server.log --endpoints endpoints.yml --debug

I’m getting this output:

ERROR    rasa.core.agent  - Could not load model due to 'ContentLength'.
UserWarning: The model at '20220605-234203.tar.gz' could not be loaded. Error: <class 'KeyError'>: 'ContentLength'
UserWarning: Agent could not be loaded with the provided configuration. Load default agent without any model.

I would really appreciate it if someone could tell me how to fetch the rasa model from S3 successfully.

Thanks.

2

Solved the issue by fetching the rasa model from S3 using Boto3 client

3

Does this mean you created a workaround? Did you ever successfully do it the normal way suggested by docs?

Because the official docs did not require rasa developer to write boto code, even if rasa may be using boto under the hood.

What caused the ContentLength error? Is it because the model is too big? If yes do you know what’s the size limit for doing it through the docs method? Have you tested whether the normal way works with a smaller model?