Being bad at a thing is the first step to being kinda good at a thing.

First, many distros ship with sudo so its pretty ubiquitous, anything you learn about managing sudo will apply to most if not all distros, not just debian. (Great choice though ❤️)

The correct answer is "it depends".

In a production environment you'll typically have some external authentication source like IdM, FreeIPA or active directory set up. In this case its common to just give full sudo access to the group that comprise your admin team, as in most cases you have to trust that they know what they're doing.

Ideally you want to follow the priciple of least access and avoid privilege escalation as much as possible. For example, there may be specific instances where a non-priv user needs to run $x as a super user, in which case, you should only grant the ability to 'sudo' for that executable as opposed to 'ALL'.

As you've already discovered, with great power comes great responsibility. 😉

Have you read their page about sudo? Gives some pretty good insight. For bonus points, at the bottom of that page are links for security and administration.

https://wiki.debian.org/sudo

There is already a sudo group. Just add a user to it and next time they log in they can use sudo.

@deepdive The ""correct"" answer is "only do root things as root" not via root. But security in more of a human problem than it is configuration.

In my home pc, I don't use sudo because my wife is the main user, and in the ultra rare occasion I need to be root in the command line (for example, if she didn't update packages from the GUI for long, I'll update but I like aptitude better), then I use su. It's a LTS 18.04 Kubuntu btw. Real users don't need root. Distro hoppers and tinkerers (nothing wrong with it) do.

On servers, I also use su. I ssh as a normal user (root ssh is usually disabled), then often immediately su, as if I'm logging into the server, it's for root work. I sometimes su - down to some specific "service" user to do that user's tasks (such as git on a gitlab server, or ndbadm on a HANA DB server).

I only tinker with sudo if I want to create users that will have one single purpose, which needs root permissions, such as restarting a service. In this case that user will be in the sudoers file, with permission for a single script or command, and often that command will be its default shell in /etc/passwd, and someone can ssh (pre shared key) to trigger it if necessary.

PS: It happened sometimes that I was given a user with full sudo permissions to do root work in someone else's server, but no root password. Then the first thing I do after logging in is "sudo su -" :-)