benz0li
commented
1 year ago I'm specifically interested in images designed for data science development environments, so general best practices for Docker and stack overflow haven't been as helpful.
While these images use s6-overlay, the Jupyter Docker Stacks images rely on tini.
@sterlinm Why are the most widely used containerised [and web-based] Data Science docker stacks using different systems for managing their processes(/services)?
👉 Choosing an init process is a matter of preference. As Dockerized Business Analytics shows, RStudio and Jupyter can also be managed with supervisord.
ℹ️ See also Choosing an init process for multi-process containers.
sterlinm
Container image name
rocker/tidyverse:4.2.3
Container image digest
No response
What operating system related to this question?
Linux
System information
No response
Question
Apologies if this isn't an appropriate question for this forum, but I've spent a lot of time reviewing these images and am trying to understand them better.
One part that I've been puzzling over for a while and have spent a lot of time reading up on now is the s6-overlay init system that is used for process supervision. I was curious whether that's still considered by the developers here to be the best solution for managing services in an image or if you'd consider using something else if you were starting from scratch.
Feel free to close this, but any advice would be welcome. I'm specifically interested in images designed for data science development environments, so general best practices for Docker and stack overflow haven't been as helpful. Thanks for your time!