minideb VS stego-toolkit

postgres:15-bullseye 2bb008a38e7c 379MB

[1] https://github.com/bitnami/minideb

However, it is sometimes a good idea to benchmark the speed of different images, as sometimes a significant speed loss is possible.

as is stated initially, that goes back to how bitnami is building its Docker images, basing on a set of debian packages (minideb) - there's also a shell library/framework embedded that does useful things, but that makes you read more code when you go check how the sausage is made. That minideb is the basis for the higher CVE count compared to scratch or alpine images.

> it’s a well-kept secret that no one wants to talk about

the maintainer side most casual docker image users aren't aware of I'd rephrase, but bitnami at least documents the issue

https://github.com/bitnami/minideb#security

https://docs.bitnami.com/kubernetes/open-cve-policy/

Ah, yeah it's a little more confusing because it's using the debootstrap tool (https://wiki.debian.org/Debootstrap) to build the container image filesystem. You can see all the gory logic here: https://github.com/bitnami/minideb/blob/master/buildone and https://github.com/bitnami/minideb/blob/master/mkimage It's a bunch of shell scripting that's not really meant to be interpreted by anyone that isn't a debian expert though, so don't feel bad if it looks really confusing. I think the overall thing is that minideb installs the absolute bare minimum system with debootstrap and even strips out a few essential packages like trusted SSL CAs, etc. If you need anything (including those essential packages) you're meant to just install_packages install them--it's all using the same apt sources and packages as debian.

Do you know why this is? Because it's part of the base file system. Here is a line from the build script for minideb (basically the smallest image needed to run a container): https://github.com/bitnami/minideb/blob/e4f37e8a5d271d93b79c3f4caa49c4ceb95d8eec/mkimage#L52

Other then that: - use `exiftool` not `exiv2` (it was able to read way more metadata last time I tried) - would definitely try `LSBSteg` and `foremost`, see more here at https://github.com/DominicBreuker/stego-toolkit - https://stegonline.georgeom.net, check out the different color masks to see if there isn't some information hidden in the color bits. Often, but not always, seeing varying pixels in one or two color channels means some kind of stego was applied or otherwise it's some custom encoding => imagemagick + python to extract)

This stego-toolkit links to a program called AudioStego in C++ that you can install with cmake.