I am using the Docker image redislabs/redismod and trying to install additional Python packages, for example Numpy, from the Dockerfile such that it can be used within a Gears.
I have tried this by following the examples provided in EdgeRealtimeVideoAnalytics and AnimalRecognitionDemo. However, these use a different base image which does not seem to be supported anymore and I was not able to apply the same approach to the redismod image.
Thanks a lot, @meirsh! But does this mean that the respective host has to have a constant connection to the internet? We need to adress situations where a connection to the internet for downloading additional packages is not available. Hence, it would be good to know whether
one is able to specify a local file system providing already downloaded Python packages
@manl we currently do not support local file system but we do have a solution, what you need to do is to spin a docker image of RedisGears in an environment that has internet and use ‘Gears-cli
install-requirements’ (https://github.com/RedisGears/gears-cli) to install whatever requirements you need. Then you can use ‘Gears-cli export-requirements’ to export the requirements you installed, it will create a zip that can then be import (again using ‘Gears-cli import-requirements’) to the environment without an internet.
Let me know if you have any issue with that.