Paranoid Penguin - Secured Remote Desktop/Application Sessions
Now that I've shown you the preferred way to run remote X Window System applications, let's discuss how to control an entire remote desktop. In the Linux/UNIX world, the most popular tool for this is Virtual Network Computing, or VNC.
Originally a project of the Olivetti Research Laboratory (which was subsequently acquired by Oracle and then AT&T before being shut down), VNC uses a protocol called Remote Frame Buffer (RFB). The original creators of VNC now maintain the application suite RealVNC, which is available in free and commercial versions, but TightVNC, UltraVNC and GNOME's vino VNC server and vinagre VNC client also are popular.
VNC's strengths are its simplicity, ubiquity and portability—it runs on many different operating systems. Because it runs over a single TCP port (usually TCP 5900), it's also firewall-friendly and easy to tunnel.
Its security, however, is somewhat problematic. VNC authentication uses a DES-based transaction that, if eavesdropped-on, is vulnerable to optimized brute-force (password-guessing) attacks. This vulnerability is exacerbated by the fact that many versions of VNC support only eight-character passwords.
Furthermore, VNC session data usually is transmitted unencrypted. Only a couple flavors of VNC support TLS encryption of RFB streams, and it isn't clear how securely TLS has been implemented even in those. Thus, an attacker using a trivially hacked VNC client may be able to reconstruct and view eavesdropped VNC streams.
Luckily, as it operates over a single TCP port, VNC is easy to tunnel through SSH, through Virtual Private Network (VPN) sessions and through TLS/SSL wrappers, such as stunnel. Let's look at a simple VNC-over-SSH example.
But Don't Real Sysadmins Stick to Terminal Sessions?
If you've read my book or my past columns, you've endured my repeated exhortations to keep the X Window System off of Internet-facing servers, or any other system on which it isn't needed, due to X's complexity and history of security vulnerabilities. So, why am I devoting an entire column to graphical remote system administration?
Don't worry. I haven't gone soft-hearted (though possibly slightly soft-headed); I stand by that advice. But, there are plenty of contexts in which you may need to administer or view things remotely besides hardened servers in Internet-facing DMZ networks.
And, not all people who need to run remote applications in those non-Internet-DMZ scenarios are advanced Linux users. Should they be forbidden from doing what they need to do until they've mastered using the vi editor and writing bash scripts? Especially given that it is possible to mitigate some of the risks associated with the X Window System and VNC?
Of course they shouldn't! Although I do encourage all Linux newcomers to embrace the command line. The day may come when Linux is a truly graphically oriented operating system like Mac OS, but for now, pretty much the entire OS is driven by configuration files in /etc (and in users' home directories), and that's unlikely to change soon.
To tunnel VNC over SSH, your remote system must be running an SSH dæmon (sshd) and a VNC server application. Your local system must have an SSH client (ssh) and a VNC client application.
Our example remote system, remotebox, already is running sshd. Suppose it also has vino, which is also known as the GNOME Remote Desktop Preferences applet (on Ubuntu systems, it's located in the System menu's Preferences section). First, presumably from remotebox's local console, you need to open that applet and enable remote desktop sharing. Figure 2 shows vino's General settings.
If you want to view only this system's remote desktop without controlling it, uncheck Allow other users to control your desktop. If you don't want to have to confirm remote connections explicitly (for example, because you want to connect to this system when it's unattended), you can uncheck the Ask you for confirmation box. Any time you leave yourself logged in to an unattended system, be sure to use a password-protected screensaver!
vino is limited in this way. Because vino is loaded only after you log in, you can use it only to connect to a fully logged-in GNOME session in progress—not, for example, to a gdm (GNOME login) prompt. Unlike vino, other versions of VNC can be invoked as needed from xinetd or inetd. That technique is out of the scope of this article, but see Resources for a link to a how-to for doing so in Ubuntu, or simply Google the string “vnc xinetd”.
Continuing with our vino example, don't close that Remote Desktop Preferences applet yet. Be sure to check the Require the user to enter this password box and to select a difficult-to-guess password. Remember, vino runs in an already-logged-in desktop session, so unless you set a password here, you'll run the risk of allowing completely unauthenticated sessions (depending on whether a password-protected screensaver is running).
If your remote system will be run logged in but unattended, you probably will want to uncheck Ask you for confirmation. Again, don't forget that locked screensaver.
We're not done setting up vino on remotebox yet. Figure 3 shows the Advanced Settings tab.
Several advanced settings are particularly noteworthy. First, you should check Only allow local connections, after which remote connections still will be possible, but only via a port-forwarder like SSH or stunnel. Second, you may want to set a custom TCP port for vino to listen on via the Use an alternative port option. In this example, I've chosen 20226. This is an instance of useful security-through-obscurity; if our other (more meaningful) security controls fail, at least we won't be running VNC on the obvious default port.
Also, you should check the box next to Lock screen on disconnect, but you probably should not check Require encryption, as SSH will provide our session encryption, and adding redundant (and not-necessarily-known-good) encryption will only increase vino's already-significant latency. If you think there's only a moderate level of risk of eavesdropping in the environment in which you want to use vino—for example, on a small, private, local-area network inaccessible from the Internet—vino's TLS implementation may be good enough for you. In that case, you may opt to check the Require encryption box and skip the SSH tunneling.
(If any of you know more about TLS in vino than I was able to divine from the Internet, please write in. During my research for this article, I found no documentation or on-line discussions of vino's TLS design details whatsoever—beyond people commenting that the soundness of TLS in vino is unknown.)
This month, I seem to be offering you more “opt-out” opportunities in my examples than usual. Perhaps I'm becoming less dogmatic with age. Regardless, let's assume you've followed my advice to forego vino's encryption in favor of SSH.
At this point, you're done with the server-side settings. You won't have to change those again. If you restart your GNOME session on remotebox, vino will restart as well, with the options you just set. The following steps, however, are necessary each time you want to initiate a VNC/SSH session.
On mylaptop (the system from which you want to connect to remotebox), open a terminal window, and type this command:
mick@mylaptop:~$ ssh -L 20226:remotebox:20226 admin-slave@remotebox
OpenSSH's -L option sets up a local port-forwarder. In this example, connections to mylaptop's TCP port 20226 will be forwarded to the same port on remotebox. The syntax for this option is “-L [localport]:[remote IP or hostname]:[remoteport]”. You can use any available local TCP port you like (higher than 1024, unless you're running SSH as root), but the remote port must correspond to the alternative port you set vino to listen on (20226 in our example), or if you didn't set an alternative port, it should be VNC's default of 5900.
That's it for the SSH session. You'll be prompted for a password and then given a bash prompt on remotebox. But, you won't need it except to enter exit when your VNC session is finished. It's time to fire up your VNC client.
vino's companion client, vinagre (also known as the GNOME Remote Desktop Viewer) is good enough for our purposes here. On Ubuntu systems, it's located in the Applications menu in the Internet section. In Figure 4, I've opened the Remote Desktop Viewer and clicked the Connect button. As you can see, rather than remotebox, I've entered localhost as the hostname. I've also entered port number 20226.

Figure 4. Using vinagre to Connect to an SSH-Forwarded Local Port
When I click the Connect button, vinagre connects to mylaptop's local TCP port 20226, which actually is being listened on by my local SSH process. SSH forwards this connection attempt through its encrypted connection to TCP 20226 on remotebox, which is being listened on by remotebox's vino process.
At this point, I'm prompted for remotebox's vino password (shown in Figure 2). On successful authentication, I'll have full access to my active desktop session on remotebox.
To end the session, I close the Remote Desktop Viewer's session, and then enter exit in my SSH session to remotebox—that's all there is to it.
This technique is easy to adapt to other versions of VNC servers and clients, and probably also to other versions of SSH—there are commercial alternatives to OpenSSH, including Windows versions. I mentioned that VNC client applications are available for a wide variety of platforms; on any such platform, you can use this technique, so long as you also have an SSH client that supports port forwarding.
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Sponsored by AMD
Built-in forensics, incident response, and security with Red Hat Enterprise Linux 6
Every security policy provides guidance and requirements for ensuring adequate protection of information and data, as well as high-level technical and administrative security requirements for a system in a given environment. Traditionally, providing security for a system focuses on the confidentiality of the information on it. However, protecting the data integrity and system and data availability is just as important. For example, when processing United States intelligence information, there are three attributes that require protection: confidentiality, integrity, and availability.
Learn more about catching the bad guy in this free white paper.
Sponsored by DLT Solutions
| Designing Electronics with Linux | May 22, 2013 |
| Dynamic DNS—an Object Lesson in Problem Solving | May 21, 2013 |
| Using Salt Stack and Vagrant for Drupal Development | May 20, 2013 |
| Making Linux and Android Get Along (It's Not as Hard as It Sounds) | May 16, 2013 |
| Drupal Is a Framework: Why Everyone Needs to Understand This | May 15, 2013 |
| Home, My Backup Data Center | May 13, 2013 |
- RSS Feeds
- Dynamic DNS—an Object Lesson in Problem Solving
- Making Linux and Android Get Along (It's Not as Hard as It Sounds)
- Designing Electronics with Linux
- Using Salt Stack and Vagrant for Drupal Development
- New Products
- A Topic for Discussion - Open Source Feature-Richness?
- Drupal Is a Framework: Why Everyone Needs to Understand This
- Validate an E-Mail Address with PHP, the Right Way
- What's the tweeting protocol?
- Kernel Problem
8 hours 29 min ago - BASH script to log IPs on public web server
12 hours 56 min ago - DynDNS
16 hours 32 min ago - Reply to comment | Linux Journal
17 hours 4 min ago - All the articles you talked
19 hours 28 min ago - All the articles you talked
19 hours 31 min ago - All the articles you talked
19 hours 33 min ago - myip
23 hours 57 min ago - Keeping track of IP address
1 day 1 hour ago - Roll your own dynamic dns
1 day 7 hours ago
Free Webinar: Hadoop
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Some of key questions to be discussed are:
- What is the “typical” Hadoop cluster and what should be installed on the different machine types?
- Why should you consider the typical workload patterns when making your hardware decisions?
- Are all microservers created equal for Hadoop deployments?
- How do I plan for expansion if I require more compute, memory, storage or networking?






Comments
Problem with conection
Thanks for this post, since it's what I needed. But I have a problem when I try the last step of the tutorial (I configured the rest exactly as described here): when I'm filling the fields as in Fig. 4 (with the same port), I receive this message (in Spanish, sorry): "Se cerró la conexión con el equipo «localhost:20226».", which means something like "Conection to localhost:20226 was closed." I don't know what can it be. Maybe the point is that I'm trying to conect Ubuntu 8.04 to 8.10 ¿? (BTW, I'm also using the "ufw" firewall and I opened 20226 ports in both computers, TCP and UDP)
Anyway, thank you in advance for any kind of help.