I have a host that I frequently ssh into. But I don't want to enter it again and again. Should I use an environment variable for this or is there a better way?

I wouldn't mind a generalized solution on saving frequently used variables that exist outside of a single bash session.

  • 3
    How about saving the params in ~/.ssh/config? Some examples are here: nerderati.com/2011/03/17/…
    – user000001
    Feb 14 '19 at 7:31
  • 4
    To expand on Roger's comment, URLs are those things that have the format of <protocol>://<host>:<port>/<path> (simplified example). The syntax ssh uses to specify locations is different. ssh ssh://user@server:22/ would be invalid.
    – JoL
    Feb 14 '19 at 20:06
  • 6
    Possible duplicate of Host alias for ssh, and this and probably a few more.
    – Sparhawk
    Feb 16 '19 at 6:22

Environment variables are for the purpose of communicating information to multiple commands/processes that you start in shell that commands or processes expect to be there in the environment. Usually such variables include options, such as LESS variable passing frequently-used options to less pager, PERL5LIB for finding perl modules in non-standard locations, LC_LANG to communicate which language a command should use for the output.

If the URL is for your own use with a specific command, use an alias such as alias au='firefox https://askubuntu.com or a function such as open_url(){ firefox "$@" } to open arbitrary URL that you provide on command-line as positional parameter to the function.

In certain cases such as ssh, you can define connection properties in configuration files as explained in Lekensteyn's answer:

  1. define ~/.ssh/config file with the following contents

    Host meh
        HostName meh.example.com
        User admin
        Port 1234
        IdentityFile ~/.ssh/id_rsa
  2. use ssh meh to connect to the host using the config file.

  • 2
    I'd also use IdentitiesOnly yes . Feb 14 '19 at 10:59
  • Also note that by default on Ubuntu, tab completion works with ssh hosts named in the config file. Feb 15 '19 at 18:57
  • 1
    @Dubu I don't understand your comment. The answer clearly contains URLs. Are you saying it doesn't? Feb 15 '19 at 19:04
  • I think the URL comment comes from referring to the connection properties in the config file as a "URL" which is not strictly accurate. I changed that wording.
    – user459652
    Feb 15 '19 at 21:47

Also I would suggest to create an alias.

Edit your .bashrc (or maybe .profile or similar file) in your home directory. Add several aliases like:

alias go='ssh url1'
alias go2='ssh url2'

Then, relogin/reconnect and enter go or go2.


I'd suggest combining Sergiy's and Lewis's answers for maximum laziness efficiency:

First create a host entry for ssh:

define ~/.ssh/config file with the following contents

    Host meh
        HostName meh.example.com
        User admin
        Port 1234
        IdentityFile ~/.ssh/id_rsa

Now ssh meh works, but that could still be quite long. There is auto-completion (after ssh[Blank]), but it's still an awful lot to type.

So let's also define an alias:

Edit your .bashrc in your local home directory. Add an alias like:

alias meh='ssh meh'

Now you can connect to "meh.example.com" by simply typing meh in your terminal window. If instead of "meh" you want to use a longer string, you can actually use [Tab] key to autocomplete.

Or if you are really lazy, just define a single character as an alias:

alias m='ssh meh'

So, if you type m and hit Enter/Return, your ssh connection will start immediately!

  • I think autocomplete for ssh-config 'just works' on ubuntu, and has worked as far as I can remember, the only place where I know it doesn't work out-of-the-box is MacOS
    – Pelle
    Feb 15 '19 at 14:37
  • @Pelle, yes that's true! I better rephrase that... But you have to type out the whole of ssh plus the blank and at least the first character.. which is so much work if you have to do it often.. and as sysadmin you should be lazy ;) Feb 15 '19 at 14:42
  • It's not clear from your answer what the benefit of combining the two approaches is, over just using the bash alias (of course I agree there is a benefit). Feb 15 '19 at 20:05
  • @Jon Bentley, hm... I'll have a look tomorrow and see how I can make this clear. Feb 15 '19 at 20:19
  • Sure. And just to elaborate on my comment a little, I'm contrasting this approach with doing a alias meh='ssh admin@meh.example.com -p 1234 -i ~/.ssh/id_rsa' which would achieve the same thing. The benefit I am thinking of of using the config file is that you can enforce the IdentitiesOnly yes option. Also that it feels more "correct" to store your SSH hosts in config rather than in some bash alias. Feb 15 '19 at 20:25

Yet another option is to use the history features of the shell. You can Ctrl-r in bash and type a unique substring of the last time you ssh'ed to that place. If there's no unique substring, you can just use whatever qualifies best as a rare substring and browse through the history entries by repeatedly pressing Ctr-r until you get to it. For example, you might type Ctrl-r and "monkeys" or "codemonkeys" or "odemon" or "kong" and get

ssh kong@codemonkeys.some-server.com

from the last, and hopefully only time you typed that whole.

With the history features of the shell, there's really no reason to type any long command more than once.


I usually just create a .sh which contains a ssh command. For example:

# file name: mywebsite-live.sh
ssh username@ # or whatever the domain is.

And then to run it I simply do ./mywebsite-live.sh

I realise that it's probably a pretty crappy solution but it works for me and the hope is that it might work for someone else too.

Not the answer you're looking for? Browse other questions tagged or ask your own question.