The major difference between sudo and su is the mechanism used to authenticate. With su the user must know the root password (which should be a closely guarded secret), while with sudo the user uses his/her own password. In order to stop all users causing mayhem, the priviliges discharged by the sudo command can, fortunately, be configured using the /etc/sudoers file.
Both commands run a command as another user, quite often root.
sudo su - works in the example you gave because the user (or a group where the user is a member) is configured in the /etc/sudoers file. That is, they are allowed to use sudo. Armed with this, they use the sudo to temporarily gain root privileges (which is default when no username is provided) and as root start another shell (su -). They now have root access without knowing root's password.
Conversely, if you don't allow the user to use sudo then they won't be able to sudo su -.
Distros generally have a group (often called wheel) whose members are allowed to use sudo to run all commands. Removing them from this group will mean that they cannot use sudo at all by default.
The line in /etc/sudoers that does this is:
## Allows people in group wheel to run all commands
%wheel ALL=(ALL) ALL
While removing users from this group would make your system more secure, it would also result in you (or other system adminstrators) being required to carry out more administrative tasks on the system on behalf of your users.
A more sensible compromise would configure sudo to give you more fine grained control of who is allowed to use sudo and who isn't, along with which commands they are allowed to use (instead of the default of all commands). For example,
## Allows members of the users group to mount and unmount the
## cdrom as root
%users ALL=/sbin/mount /mnt/cdrom, /sbin/umount /mnt/cdrom
(only useful with the previous %wheel line commented out, or no users in the wheel group).
Presumably, distros don't come with this finer grained configuration as standard as it's impossible to forecast what the admin's requirements are for his/her users and system.
Bottom line is - learn the details of sudo and you can stop sudo su - while allowing other commands that don't give the user root shell access or access to commands that can change other users' files. You should give serious consideration to who you allow to use sudo and to what level.
WARNING: Always use the visudo command to edit the sudoers file as it checks your edits for you and tries to save you from the embarrassing situation where a misconfigured file (due to a syntax error) stops you from using sudo to edit any errors. This is especially true on Debian/Ubuntu and variants where the root account is disabled by default.
The point of the nologin shell is to prevent the user from logging in. Such a user may still use your server services like FTP, IMAP/POP3 and others but they won't be able to login e.g. using sshd or console, period.
How do I switch from root to a user with a nologin shell?
sudo -u USERNAME /bin/bash
Will work but only root can do that.
Best Answer
su(mostly) uses pam for authentication and pam has a module called pam_wheel which checks group membership of the authenticating user. In short, by addingto the file
/etc/pam.d/su, only users who are members of the groupbecomerootmay become root using su. Now you make sure only your user EMERG is a member of the group becomeroot. Some distros have/use the group namedwheelfor that.Further reading: pam (7) pam_wheel (8) groupadd (8) gpasswd (1) and many distros have explaining comments in
/etc/pam.d/suas well