this post was submitted on 20 Dec 2023
45 points (97.9% liked)

Linux

48209 readers
751 users here now

From Wikipedia, the free encyclopedia

Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).

Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.

Rules

Related Communities

Community icon by Alpár-Etele Méder, licensed under CC BY 3.0

founded 5 years ago
MODERATORS
 

I don't like my ssh keys being stored in plain sight, I also don't like having to type a passphrase to use them.

On windows, once you run ssh-add, the key is stored in a secure way and managed by some kind of session manager (source), at that point you can delete the key file and go about your life knowing that the key is safe and you won't need to type a password again.

I would like something similar on linux, like storing the key via libsecret as you do with git, so that you can access your servers without having a key in plain text.

I think it's possible to generate a key with a passphrase and have gnome-keyring or kwallet remember the passphrase, but it would be nicer to just securely store the key itself.

Can that be done?

you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 10 points 11 months ago (1 children)

Yes, it can be done. Not to the point of deleting your key (that makes no sense - you need the key), but ssh-agent is what you want. Add it to your shell config and it will only ask to be unlocked once per however often you define.

I have this function defined and called:

function ssh-agent-setup() {
    # SSH agent
    pid_file="$HOME/.ssh/ssh-agent.pid"
    SSH_AUTH_SOCK="$HOME/.ssh/ssh-agent.sock"
    if [ -z "$SSH_AGENT_PID" ]
    then
      # no PID exported, try to get it from pidfile
      SSH_AGENT_PID=$(cat "$pid_file")
    fi

    if ! kill -0 "$SSH_AGENT_PID" &> /dev/null
    then
      # the agent is not running, start it
      rm "$SSH_AUTH_SOCK" &> /dev/null
      >&2 echo "Starting SSH agent, since it's not running; this can take a moment"
      eval "$(ssh-agent -s -a "$SSH_AUTH_SOCK")"
      echo "$SSH_AGENT_PID" > "$pid_file"

      >&2 echo "Started ssh-agent with '$SSH_AUTH_SOCK'"
    fi
    export SSH_AGENT_PID
    export SSH_AUTH_SOCK
}

ssh-agent-setup

This way it stores the unlocked key in memory until the end of the session.

[–] [email protected] 6 points 11 months ago (2 children)

that makes no sense - you need the key

But if it's stored in a keyring or similar (like on windows) and the client reads from it you don't need the file with the plain text key. Like you don't store the git credentials in a file, but with libsecret.

I would prefer something that never ask for the password.

Things like the gnome-keyring or kwallet keep all the passwords in an encrypted file, they get decrypted and kept in ram using your login password when you log into gnome/KDE session and programs can ask for passwords using some API. Once you log out the passwords are removed from ram and no one can read them. My goal is to have something like this, so I'm never asked for a password, I just log into my session and everything is available

[–] [email protected] 3 points 11 months ago

I don't like what you're trying to do, but I think gnome-keyring would do this for you. Seahorse is the gui for it

[–] [email protected] 1 points 11 months ago

I'n Windows it is not stored in a keyring but instead in the registry. This has basically the same security threat model as a local key file.

The ssh-agent on Linux will do what you want with effectively the same security. The biggest difference being that it doesn't run as a system service but instead runs in userspace which can make it easier to dump memory. There are some other agent services out there with additional security options but they don't change the threat model much.