If you run multiple containers of the same image, or similar images with similar core/base, does each container run its own instance/copy of the necessary core/base functions?

[imthenachoman]

Say I have an image that opens a shell, writes the current date to a file, sleeps 60 seconds, and loops.

If I run multiple instances of the image, multiple containers, does that mean my system is running multiple instances of the shell and my script?

I ask because I see a lot of images for light-weight services – but those light-weight services need the same/similar core/base functions. In those cases, wouldn’t it make more sense for me to create my own image with all of the light-weight services – all using the same core/base functions? Wouldn’t that be more efficient.

And yes, I know it is painful to create an image that runs multiple services – but it is possible.

[meyay (Metin Y.)]

@imthenachoman

If I run multiple instances of the image, multiple containers, does that mean my system is running multiple instances of the shell and my script?

Absolutly!

@imthenachoman

I ask because I see a lot of images for light-weight services – but those light-weight services need the same/similar core/base functions. In those cases, wouldn’t it make more sense for me to create my own image with all of the light-weight services – all using the same core/base functions? Wouldn’t that be more efficient.

Yep. DRY is always a good idea Depending on the design of your images, you might still end up running multiple instanxes of your script…

@imthenachoman

And yes, I know it is painful to create an image that runs multiple services – but it is possible.

Just because you have the freedom to shoot yourself in the foot, doesn’t mean that you’ll like the outcome. Just don’t! … that is of course unless there is an inevitable reason to do so…

[imthenachoman]

So I guess there is a balance to be had. I might try to create my own image for Plex and a few related services – just for learning purposes.

Thank you!

[meyay (Metin Y.)]

Just out of curriousity: what improvements do you have in mind?

Also, if you adopt the official PMS image, be aware that they use s6-overlay as init system, which follows a convention where files need to be located in your image.

[imthenachoman]

@meyay

Just out of curriousity: what improvements do you have in mind?

Well, I don’t know yet. My thought is, Sonarr, Radarr and SabNZBD are all rather light-weight – idle 95% of the day. Instead of having 4 containers for them (including Plex), why not put it in one image. I mean, I don’t know if the performance cost will be worth it but I’m game to try it out and see. Plus, that way I’ll better learn Docker.

[meyay (Metin Y.)]

Appart from have a fun exercise, I am not sure if you actualy gain any benefit from the endavor - remember: only the required processes are started and running. Though, on the other side you create yourself a harder maintainable beast of an image. Seperation of concerns realy is a thing

I feel the time spend to work on clever compose-files or even learn kubernets is effectivly better spend then creating your personal maintanance hell.