ssh Command in Linux: Work on Remote Machines

ssh opens a shell on another machine over an encrypted connection. It is how nearly all server work happens, and its config file quietly turns twenty-character commands into two-letter ones.

Connect #

Connect to a remote machine

$ ssh ana@203.0.113.7
$ ssh -p 2222 ana@build.aospix.com

The first connection asks you to trust the host key — say yes from a machine you control — then authenticates you. Passwords work, but keys are better: ssh-keygen creates a pair once, and ssh-copy-id ana@host installs your public key so logins after that need no password at all.

Run one command and leave #

Everything after the host executes remotely, and the connection closes when it finishes:

Check disk space on a server without logging in

$ ssh ana@build.aospix.com "df -h /"

That form is half of every automation script; the other half is curl when the remote thing speaks HTTP.

The config file #

A ~/.ssh/config entry turns hosts into names:

A ~/.ssh/config entry

Host build
  HostName build.aospix.com
  User ana
  Port 2222
  IdentityFile ~/.ssh/id_ed25519

After that, ssh build carries all four settings — and scp and git remotes inherit the same shortcut.

The flags worth knowing #

Flag	Meaning
`-p port`	Connect to a non-default port
`-i file`	Use a specific identity (key) file
`-v`	Verbose — the debugging lifeline when login fails
`-L 8080:localhost:80`	Forward a local port to the remote side

Once logged in, everything in this series applies remotely: systemctl for services, ps and top for load, df for disks. More in the journal.