Hey Jonathan, great article and to the point! Since I, myself, is a noob in the Docker realm, I just want to clarify something: with this method we are not building our own containers from our Rasa installation, rather than pulling images from dockerhub, right?
Newbie Data Scientist (Still) / Machine Learning / Natural Language Processing / Rasa Certified Developer, Enthusiast and Rasa Hero / Conversational Agent Psychologist ;-)
Sorry for the crazy delay in answering, I'm sure you figured it out by now. I didn't get any notification there was a reply and was just clicking around and saw I had 2 comments here.
Anyway - You're correct, this procedure pulls down the Rasa containers from Docker Hub. It makes a quick win, however, the rasa-sdk image for the action server is HUGE and if you wanted to speed up the process, you could create your own file and spin up a custom "sdk" container for the action server. I know when we build/deply these for our dev environment they're close to 2G - which is beyond huge for a container.
Unfortunately I've gotten too lax with "it just works", but really should dive in and rip out all the extra stuff I'm not using to see if it makes any difference. There are a lot of packages and channels configured in the DockerHub containers that could be stripped out to make them a bit leaner.
That may be a good idea for another article - streamlining your docker containers.
Good luck with your project, let me know if you dove into creating your own, I'd love to chat about how you did that and what you ripped out.
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
Hey Jonathan, great article and to the point! Since I, myself, is a noob in the Docker realm, I just want to clarify something: with this method we are not building our own containers from our Rasa installation, rather than pulling images from dockerhub, right?
Sorry for the crazy delay in answering, I'm sure you figured it out by now. I didn't get any notification there was a reply and was just clicking around and saw I had 2 comments here.
Anyway - You're correct, this procedure pulls down the Rasa containers from Docker Hub. It makes a quick win, however, the rasa-sdk image for the action server is HUGE and if you wanted to speed up the process, you could create your own file and spin up a custom "sdk" container for the action server. I know when we build/deply these for our dev environment they're close to 2G - which is beyond huge for a container.
Unfortunately I've gotten too lax with "it just works", but really should dive in and rip out all the extra stuff I'm not using to see if it makes any difference. There are a lot of packages and channels configured in the DockerHub containers that could be stripped out to make them a bit leaner.
That may be a good idea for another article - streamlining your docker containers.
Good luck with your project, let me know if you dove into creating your own, I'd love to chat about how you did that and what you ripped out.