Browse securely from a public WiFi connection with SSH

If you are on the road and jump on your hotel's free WiFi be afraid.  Be very afraid.  Why?  All of your network traffic is being broadcasted in all directions with no security protection whatsoever, and it is insanely easy for anyone to read that traffic using freely available tools.  Using ARP spoofing it is possible for a snooper to associate his MAC address with the network's gateway, thereby routing all internet-bound traffic through his or her machine.  Even SSL is not entirely safe as an attacker can use SSL's renegotiation capability to trick a server into giving the attacker access to your session.

What's a Linux geek to do?  Simple: SSH!

Using SSH you can create a secure tunnel from your laptop to your home computer, and pass all of your web traffic coming from your laptop through that connection to your home computer, then out to the internet.  It's simple; here's how it's done:

1. Before you leave your home, make sure your SSH server is set up properly (see my instructions on how to harden it).  We'll say for simplicity that you have configured your SSH server and all of your clients to run on port 1234.  That means that your SSH server is listening for incoming connections on port 1234.
2. Log into your router's configuration interface and configure it to forward all incoming traffic that arrives at port 1234 to your SSH server's IP address.
3. Go to www.whatismyip.com and write down your public IP address.  For our example, we'll say that it is 200.200.200.200.
4. When you arrive at your hotel, coffee shop, library, etc., pick a port above 1024 to use on your laptop.  The SSH client will listen on that port and forward all traffic to your home machine; for our example it will be port 4321. 
5. Open a terminal and run the following command: ssh -D portNumber -N publicIPAddress (in our example, it would be ssh -D 4321 -N 200.200.200.200).  If your SSH client is not already configured to use your custom port, add "-p 1234".  After you hit Enter, your cursor will move to the beginning of the next line and sit there blinking.  This means that the tunnel has been established and it is ready to start forwarding traffic.  Leave the terminal open.
6. Go into Firefox and go to Edit, then Preferences.  When the Preferences menu pops up, click the Advanced icon at the top, flip to the Network tab and hit Settings at the top of the screen.  That will pop up yet another window.  Put the bullet in "Manual Proxy Configuration" and in the box next to "SOCKS Host" enter "localhost", and enter the port you selected in step 4 (in our example, 4321) into the Port box.  Below is a screenshot of what it should look like.
7. Hit OK, then Close, and restart Firefox.
8. Browse to www.whatismyip.com and verify that it detects your home IP address as the one you are browsing from.  If whatismyip.com display's your home IP address, you have successfully configured Firefox to tunnel through SSH!

 Now you can browse in peace, and when you are finished simply flip back over to your terminal window and do a Ctrl + C to stop the SSH client.  Good luck!

Follow-up to "Using passphrase-protected SSH keys in Cron"

So I wrote a while back about automating SSH jobs in Cron when the keys were passphrase-protected.  I was very excited about being able to automate tasks over SSH and I was sure I had it working but a few days later I realized that it wasn't working at all.  It was failing without any error I could see.

I'll spare you the gory details of how many hours I spent and how many things I tried when getting it working.  The bottom line is that this evening I realized that when my Cron job or script ran "ssh-agent -s" another ssh-agent process was being created!  I figured that that command was simply exporting the environment variables necessary for me to use the process created by the Gnome session (you may remember that I have seahorse set to automatically unlock my ssh key when I log in), but instead I was inadvertently creating a new ssh-agent process with no keys in it.  Then when I tried to run Unison, it found the wrong ssh-agent process.

To fix it, I needed to find a way to keep from starting another ssh-agent process, and instead gain access to the one seahorse starts every time I log in.  I made a change to my crontab that would search for the existing ssh-agent process ID and authentication socket and import them into the cron environment.  This is kind of a hack but it actually works (not like last time).  Just add the following to your script before trying to connect to your SSH server (or do what I did and put them right into the cron job, separated by semicolons):

    export SSH_AGENT_PID=`ps -a | grep ssh-agent | grep -o -e [0-9][0-9][0-9][0-9]`
    export SSH_AUTH_SOCK=`find /tmp/ -path '*keyring-*' -name '*ssh*' -print 2>/dev/null`

XRDP, the best remote access software under Linux

My desktop PC is in my basement office, which is a nice place but somewhat inconvenient.  I really wanted a way to access it remotely from my laptop but had trouble finding a remote-access application that was fast enough.

Both my laptop and desktop are wireless, so there are 2 wireless "hops" between my machines and when logged into my desktop from my laptop each command (i.e. a click or keystroke) has 4 wireless hops to go through before my screen displays the result.  So speed is key. 

Remote access servers that didn't work out

I tried pretty much every remote access server I could find.  I was already familiar with VNC, and that was a perfectly viable option but my desktop has dual monitors, so there was a pretty big difference in screen resolutions and one of VNC's real weaknesses IMHO is that the client is forced to use the screen resolution of the server.  I didn't want to be scrolling around a lot so I axed that idea. 

Next I tried SSH with X forwarding enabled (using the -Y option).  That was nice and simple since all of the software and keys were already in place, but it was too slow, even with the -C option (which enables compression, effectively speeding up the rate of transfer).  FreeNX is a similar solution, and although it was somewhat faster some of the time, it didn't provide a consistently high level of performance so I had to reject both solutions.


Gnome also offers the option of logging into another machine via XDMCP.  When booting my laptop, I would stop at the login screen, hit Options, then "Remote Login via XDMCP".  In addition to stopping me from running any applications or doing anything else on my laptop while logged into my desktop remotely, this proved to be too slow also, so I canned the idea.


I was about to give up when I discovered xrdp.


XRDP

XRDP is a service that I installed on my desktop to enable it to accept RDP connections from my laptop.  Ubuntu already comes with an RDP client, tsclient, installed by default, and I've used Windows XP's Remote Desktop a lot at work so I know how well it works.  XRDP is in the Ubuntu repositories so I installed it on my desktop and now I can log into it from my laptop using the same program I use to log into my computer at work, and the speed more than adequate.  It is really great to be able to program, run virtual machines, and do many other resource-intensive operations from my laptop.

XRDP has a few quirks that I decided I could live with, but it is worth mentioning them here:

1. It depends upon the package "tightvncserver" being installed on the server (the machine you are logging into), although it isn't configured as a dependency in the package (this was the case as of the last time I set up XRDP, which was a while ago so hopefully they have fixed this).  Install tightvncserver before you attempt to install XRDP.
2. You may install the XRDP package and have it work just fine until you reboot the server.  This is because XRDP sometimes has difficulty figuring out whether it is running or not, so when the server is rebooting and it tries to start XRDP, XRDP thinks it is already running even though it isn't and refuses to start.  You can fix that by running this:
        sudo sesman --kill
        sudo xrdp --kill
        sudo rm /var/run/xrdp/xrdp.pid
        sudo sesman
        sudo xrdp
   
3. There is no way to log into the console session.  So if I was working at my desktop and had to get up for something, I can't resume what I was doing on my laptop.  Windows XP provides the "/console" command-line switch which allows you to take over the session that is currently displayed on the machine's screen, but there is no way to do that with XRDP.  There might be a way to do it within Gnome; I'm not sure at the moment.
5. If you do anything in the remote session that makes a sound, the sound will be played on the server's speakers, not your client's speakers.  So the first thing I hear when logging into my desktop is the sound of the Ubuntu login music being played on my desktop speakers, down the stairs and two rooms away.  This is a very minor annoyance to me but it is worth mentioning.

So now I have exactly what I set out to get: an easy-to-use remote access method that is fast and reliable.  It is really great to be able to open tsclient (you will find it on the Internet menu, it is called "Terminal Server Client") and see my desktop.  I highly recommend it!

Using passphrase-protected SSH keys in Cron

Just a few days ago I told you all about the glory that is SSH, and now I want to show you how to automate its use.

I'm sure you noticed while you were setting up your key that it asked you for a passphrase to protect the key.  If you opt not to set a passphrase, it will store your key in plain text, which means that if anyone gets access to your filesystem, they have access to your key.  They can copy it and use it on any computer to log into your server as you, so you are much better off protecting your key with a passphrase.  The key generator will use the passphrase to encrypt your key and give you protection from this threat.

Because the SSH key is encrypted, whenever you want to use it, it will ask you for your passphrase so that it can get access to the key.  This is exactly what you want to keep your key secure, but if you want to automate any of your SSH related tasks in Cron, you will be prompted for the passphrase every time SSH tries to establish a connection with your server.  One quick, easy way out of this situation is to simply keep your SSH key unencrypted and hope that no one who gets near your computer has malicious intentions, but there's a better way: ssh-agent.

Ssh-agent is a process that stores SSH keys in memory.  When you set up your SSH key with a passphrase, you may notice that Gnome gives you the option to "unlock this key automatically when I log in" (Ubuntu offers this option, YMMV); if you select that option, it will use Ssh-agent to store your SSH key in memory.

So at this point you have your SSH key encrypted on disk but unlocked in memory, so you can manually SSH into your server without entering the password for you account on the server or your passphrase for the SSH key, but you want Cron to have the same ability.  At this point, any SSH jobs in Cron will prompt you for your passphrase to unlock the SSH key.  I will show you how to fix that by making your unlocked SSH key (which is stored in memory by ssh-agent) accessible to Cron.

There are 2 steps involved in making your key available to your script running in the Cron environment: Saving the 2 SSH environment variables to disk and reading them into your environment within your Cron job.

Saving the SSH environment variables

Go into your crontab file by entering "crontab -e" at the command line.  Add the following on its own line:

     @reboot ssh-agent -s | grep -v echo > $HOME/.ssh-agent

The @reboot is a special token in cron that will cause the command to be run when you first log in.  The command will cause the SSH_AUTH_SOCK and SSH_AGENT_PID to be written to a file on disk.

Retrieving the SSH environment variables within a Cron job

Basically, in order to make the running instance of ssh-agent available to your cron job, the following line needs to be run:
      
     . ~/.ssh-agent

On my machine, I put it right into crontab just before the job I wanted to run, like this:

     # m h  dom mon dow   command
     30 * * * * . ~/.ssh-agent; unison default -ui text -batch

If you are running a script, you can alternatively put the command at the top of the script.

And that's all there is to it!  You should now be able to run an SSH command using an SSH key that is encrypted on disk but unencrypted in memory within a cron job. 

EDIT:  A few days after posting this, I discovered that it was actually not working for me at all.  The steps I just described probably would have worked just fine if I was trying to start my own ssh-agent process and manually add keys to it using ssh-add, but I was trying to do it using the ssh-agent process and key which seahorse sets up for me when I first log in, and this configuration required a slightly different solution.  See my follow-up post for details.
 

See also:  http://sial.org/howto/openssh/publickey-auth/

SSH, the Secure Shizzle

SSH, the "secure shell" is the ultimate exception to the "security is inversely proportional to convenience" rule.  It is so easy to set up and once you have it running, secure authentication and communication are totally transparent.  It's almost irresponsible not to use it.

In its most basic form, SSH is a secure remote console program, a secure replacement for Telnet, however, it is very flexible and can handle such tasks as secure file copying between machines, proxying, and even tunneling.  Also, it's been around so long that it has become one of the de facto standards for secure authentication and communication on Linux.  Pretty much any network-enabled program worth its salt will feature support for it.

There are many fine tutorials on how to use SSH, my favorite is by the incomparable Chess Griffen on the Linux Reality podcast.  If you aren't already familiar with SSH, stop reading this post and listen to that episode of Linux Reality right away.  I'll wait.

As awesome as it is to be able to type "ssh myserver", enter the password, and start entering commands to be run on your server, there's a better way.  It's called SSH keys, and they will save you the trouble of typing in your password every time you want to connect to the server and they will give your server extra protection from being hacked.  It's easier and more secure, which almost never happens, so you know you have to do it.

Creating an SSH key

Simply hop into the command line and type "ssh-keygen".  It will ask you to give a filename and location for the new keys, and it will ask you to give it a passphrase.  When it is finished, go into your home directory's .ssh folder, and you will see two files: id_rsa and id_rsa.pub.  The id_rsa file is your private key; leave it where it is and don't let anyone have access to it.  Copy the id_rsa.pub file to your server.  On the server, go into your home directory's .ssh folder.  Create a file called authorized_keys if it doesn't already exist and copy the contents of your id_rsa.pub file into it (append the contents if the authorized_keys file already exists).  Now, go back to your client machine and SSH into your server.  It should let you in without asking for a password.

More SSH hardening measures

Let's face it, half the point of SSH is the security.  Setting up keys as I described will protect communications between your computers and ensure that you are actually contacting the server you think you are contacting with almost no effort on your part.  There are a few extra steps that should be taken to gain the highest level of security:

1. Disallow protocol 1: On the server, open /etc/ssh/sshd_config as root.  Find the line that says "Protocol 1,2".  Delete the 1 and the comma, so it only says "Protocol 2".  Don't close sshd_config yet...
2. Also in sshd_config, find the line that says "#PasswordAuthentication yes".  Remove the hash mark from the beginning of the line and change "yes" to "no".  Keep sshd_config open...
3. Find the line that says "Port 22" and change the 22 to another number between 1024 and 65535.  
4. Save that file and restart the ssh daemon (you can usually do this by running "sh /etc/init.d/sshd restart" as root)
5. On your client, open /etc/ssh/ssh_config as root
6. Find the line that says "Port 22" and change it to the same port number you set in step 3 above.
7. Find the line that says "Protocol 1,2" and change it to "Protocol 2", as you did on the server in step 1 above.
8. Find the line that says "Cipher aes128-cbc,3des-cbc,blowfish-cbc...".  Remove the hash mark at the beginning of the line if there is one, then remove all of the ciphers listed except aes256-cbc.  So that line should read, "Cipher aes256-cbc".
9. Save that file
10. Bookmark this blog post, because you will have to repeat steps 5-9 on all of your clients.

This may seem like a lot of work, but please keep in mind that SSH has been around for a very long time and there are a lot of automated ways to attack an SSH server that rely on the default settings, such as support for Protocol 1 and having the SSH daemon listen on port 22.  Changing these defaults takes only a few minutes but makes it exponentially harder for hackers to get in.  These steps are absolutely essential if you are considering exposing your SSH daemon to the public Internet.  

At this point, you are able communicate between machines with more ease and convenience than ever, and those communications are secured by military-grade authentication and encryption.  Security and convenience will never be this close, so take advantage!

Unify your files with Unison

If you are using multiple computers, you must have some way of keeping the files you want on each machine.  A lot of people use Dropbox or LiveMesh, and Canonical recently threw its hat into the ring with UbuntuOne, but all of these solutions provide a pretty small amount of storage (2 gigs in the case of Dropbox and UbuntuOne), and as with any online storage solution they also require that you trust them not to lose your data and to respect your privacy.

Because I prefer not to trust anyone with my data unless I have to, and partly because I'm a cheapskate, I prefer the DIY approach.  This is where Unison comes in.

Unison Explained

Unison is a file synchronization utility that keeps the contents of two directories synchronized.  In true Unix fasion, it doesn't reinvent the synchronization wheel, instead it uses rsync to do the comparison and synchronization, effectively making a tool that could only handle one-way synchronizations capable of handling two-way synchronizations.  In addition to being able to sync local directories, it can also operate over SSH, so you don't have to spend a bunch of time worrying about providing unison a protocol or dedicated port, or setting up a dedicated authentication mechanism.

Like all great Linux utilities, Unison has a graphic user interface but can also be used on the command line (this is important for automating synchronizations in cron; more on that later).  Either way, you are going to manage Unison tasks with Profiles.  Profiles are basically a set of two folders that you are syncing (one local, the other local or remote), and zero or more options, and although they can get rather elaborate, don't let it scare you.  The Unison GUI makes the task of creating a new profile simple.

Creating Unison Profiles
  
After you've installed Unison (on Ubuntu use "sudo apt-get install unison unison-gtk"), fire it up from your menu or by typing "unison-gtk" into the command line.  You will be greeted with a screen where you can type in the name of your first "root", or local directory.  When you are finished with that, hit OK and then you will be promted to enter the second directory.  This screen will give you the option to connect to a remote machine over SSH or a raw socket (not recommended).  If you decide to sync with another machine over SSH, be sure to type in the absolute path to the folder on that machine (i.e. "/home/jizldrangs/documents").  Fill in the host name, username to connect as, and port name as necessary, then hit OK and you're done!  Unison will take this information and create a new profile called "default".  You can create new profiles the next time you launch the Unison GUI

The GUI will help you set up a profile and will allow you to set some of the basic options, but if you want to use any of the more advanced features (see the Unison man page for a list), you will need to edit the profile by hand.  Your profiles are stored in separate files in the .unison directory in your home directory (e.g. /home/jizldrangs/.unison; it is a hidden directory).  In that folder you will see a file for each of your profiles, all ending with ".prf".  Simply open the profile you want and start adding options.  You can do things like exclude certain file types, exclude certain directories, exclude files over a certain size, have it follow links, etc. 

I already have an old Pentium 3 laptop recommissioned as our file and print server, and it runs all the time, so it was the perfect place to store the master copies of all my files, and to receive updates from whichever of my computers I make changes on and distribute them to the rest of the machines, all within the security and ease of SSH.

To make sure that I have the latest version of my files, I've tasked Cron, my personal Linux butler, with the task of running the sync at the bottom of every hour.  Here is the relevant line from my crontab file:

# m h  dom mon dow   command
30 * * * * unison default -ui text -batch

The "ui text" option tells unison to use command-line mode and not to launch the gui, and the "batch" option tells it to accept default update options (basically, replace older files with newer files), so it doesn't prompt me for input on what to do with the files.  Now my file syncing is totally automated.

Unison Gotchas

If you are syncing files to an mp3 player or usb flash drive, add the option "perms = 0" to avoid getting the error "failed to set permissions" when you try to sync.

Happy syncing!  

Great command-line tools and tricks

This was on Hacker News this morning.  It is easily the best list of little-known command line tools and tricks I've seen.  Did you know that you can control MPlayer remotely from the command line?  Or that you can establish an SSH connection with a remote server, go to the server, and use the connection to access the client?  If not, you should check this out. 

http://www.tuxradar.com/content/command-line-tricks-smart-geeks