SSH trust and keys

November 10, 2019 - Reading time: 5 minutes


In this post I'll show practical examples on how to verify that we are connecting to the correct SSH endpoint and how to exchange keys for password-less authentication.

Who are you?

When we try to connect to an SSH server for the first time, we get a warning about the authenticity of the host we are about to connect, followed by a fingerprint and a question.. but where does the signature comes from? and how can we be certain that it comes from the right host?

root@chromeos:~# ssh
The authenticity of host ' (' can't be established.
ECDSA key fingerprint is SHA256:DqEhLL8M8YjVJk8XDIdlOJQ235faGOjkL8MmZ+yPxWM.
Are you sure you want to continue connecting (yes/no)? 

In older versions of the ssh client we used to get an MD5 fingerprint

root@chromeos:~# ssh -o FingerprintHash=md5
The authenticity of host ' (' can't be established.
ECDSA key fingerprint is MD5:8a:4a:b9:84:81:f8:0f:de:cb:89:be:85:42:1a:5c:30.
Are you sure you want to continue connecting (yes/no)?

In this example we are getting the ECDSA (Elliptic Curve Digital Signature Algorithm) fingerprint, but we could specify a different key algorithm

root@chromeos:~# ssh -o HostKeyAlgorithms=ssh-rsa
The authenticity of host ' (' can't be established.
RSA key fingerprint is SHA256:4skDV2xIXiQLwe0EsAdCmUjXm8CteDN2DGQcrx61lQk.
Are you sure you want to continue connecting (yes/no)?

A full list of supported key algorithms is available by running the following command:

root@chromeos:~# ssh -Q key

These fingerprints are generated in the client, from the public keys in the server. In this case, the server has:

root@raspberrypi:/etc/ssh# ll ssh_host_*pub
-rw-r--r-- 1 root root 606 Sep 26 01:24
-rw-r--r-- 1 root root 178 Sep 26 01:24
-rw-r--r-- 1 root root  98 Sep 26 01:24
-rw-r--r-- 1 root root 398 Sep 26 01:24

To verify that we are connecting to the right server, we need to compare the public key in the server and the one that we are receiving in the client.

root@raspberrypi:/etc/ssh# cat 
ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBFfsub7U+Sb6h4obExzsxzXanqzVfE4sNozlkl2Bu0j/f5ZomIv5ieY+5oTZvutW/e/eLrZ6LL5McNJJ8WR2eIo= root@raspberrypi

Then we can get the full key in the client

root@chromeos:~# ssh-keyscan -t ecdsa 
# SSH-2.0-OpenSSH_7.9p1 Raspbian-10+deb10u1 ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBFfsub7U+Sb6h4obExzsxzXanqzVfE4sNozlkl2Bu0j/f5ZomIv5ieY+5oTZvutW/e/eLrZ6LL5McNJJ8WR2eIo=

And finally, to go full circle, and check that the digest of the key is the same as the one we get in the question, we can use key-gen

root@chromeos:~# ssh-keyscan -t ecdsa | ssh-keygen -lf -
# SSH-2.0-OpenSSH_7.9p1 Raspbian-10+deb10u1
256 SHA256:DqEhLL8M8YjVJk8XDIdlOJQ235faGOjkL8MmZ+yPxWM (ECDSA)

I trust you... now what?

Your client keeps a list of all the hosts where the authenticity has been verified. When we answer the question with yes, we add this new host to the list.

root@chromeos:~# tail -n1 .ssh/known_hosts
|1|ZCjQ5DVz31CIjAcqJlhL6/PBce0=|1Bzfnbf0BUBYmsmSYHXtmIICRqY= ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBFfsub7U+Sb6h4obExzsxzXanqzVfE4sNozlkl2Bu0j/f5ZomIv5ieY+5oTZvutW/e/eLrZ6LL5McNJJ8WR2eIo=

We can see the public ECDSA key of the server in that output. But what are those strings that precede the key? This is the result of an ssh option called HashKnownHosts. When the option is enabled, instead of storing the IP address or hostname, we take this value, add some salt to it, and then hash it. So, what we have is:

HASH_MAGIC Salt Hashed value
1 ZCjQ5DVz31CIjAcqJlhL6/PBce0= 1Bzfnbf0BUBYmsmSYHXtmIICRqY=

Rely on key authentication

Before moving to this step, make sure you have a backup of your private key, and if possible a "paper key" version. Change the following values on the sshd config, and restart the service:

sudo sed -i  's/PasswordAuthentication.*/PasswordAuthentication no/g' /etc/ssh/sshd_config
sudo sed -i  's/ChallengeResponseAuthentication.*/ChallengeResponseAuthentication no/g' /etc/ssh/sshd_config

SSH port forwarding via tunneling

July 26, 2020 - Reading time: 2 minutes

Use this to create an encrypted connection between a local/remote box.

*** Local forwarding

Forwards a port from the client machine to the server machine. The ssh client listens for connections on a configured port, tunnels the connection to the server, and the server then connects to the destination address/port.

For example, I want that every client connecting to my IP address on port 8080 to be tunneled to my server on and then have the connection sent to on port 80:

root@twickenham:~# ssh -L's password:

Now we can see that the ssh client is listening on port 8080 and also has an established connectin, the tunnel, to the server:

root@twickenham:/home/ptosiani# lsof -P | grep 3703 | grep TCP
ssh       3703                      root    3u     IPv4              40524       0t0        TCP> (ESTABLISHED)
ssh       3703                      root    4u     IPv4              35436       0t0        TCP (LISTEN)

This is what we see in a packet capture when a client on the network connects to

17:49:47.131557 IP > Flags [S], seq 697753369, win 64240, options [mss 1460,nop,wscale 8,nop,nop,sackOK], length 0
GET / HTTP/1.1
Connection: keep-alive
17:49:47.133320 IP > Flags [P.], seq 2537745498:2537745598, ack 3952728736, win 501, options [nop,nop,TS val 3200931444 ecr 4168325949], length 100
# Then on the server 
16:49:47.160112 IP > Flags [P.], seq 1:508, ack 1, win 502, length 507: HTTP: GET / HTTP/1.1
E..#:.@.@.......t.7j...Pm5h..[5;P.......GET / HTTP/1.1
Connection: keep-alive
Cache-Control: max-age=0
# Response on the client shows the IP addrss of the server 
HTTP/1.1 200 OK
Server: nginx
Date: Sun, 26 Jul 2020 16:49:47 GMT
Content-Type: image/
Content-Length: 16
Connection: close
X-SECURITY: This site DOES NOT distribute malware. Get the facts.
X-RTFM: Learn about this site at and do not abuse the service.
Access-Control-Allow-Origin: *
Access-Control-Allow-Methods: GET

Make and display HTML from md

November 24, 2021 - Reading time: ~1 minute

Markdown is a basic markup language used to format text written using a text editor.
Here is a useful cheat-sheet I'm using while writing this post: cheat-sheet

To convert the .md file into .html I'm using showdown, specifically the CLI tool.
Once installed, create the following function on your ~/.bashrc:

# Function to create html from markdown 
# requires
makehtml() { 
  \rm out.html ; 
  showdown makehtml -mi $1 -o out.html ; 
  echo "<style>code { color : darkblue; background-color : beige ; } html { width: 60%;margin-left: 100px;margin-right: 100px; }</style>" >> out.html
  firefox-esr ./out.html