<metaname="description"content="Turn the Beaglebone Black into a personal communications server"
/>
<metaname="keywords"content="freedombox, debian, beaglebone, red matrix, email, web server, home server, internet, censorship, surveillance, social network, irc, jabber"/>
<h2id="unnumbered-1">Why not supply a disk image download?</h2>
<divclass="outline-text-2"id="text-unnumbered-1">
<p>
Shipping a Freedombone disk image ready to install on a flash disk would be easy, but disk images are relatively opaque. It would be quite easy to hide something nasty within a disk image and the user might never know. To guard against that possibility installing via the <i>install-freedombone.sh</i> script is a lot more transparent. You can check the code to see exactly what it's doing, and the packages are all downloaded from standard Debian repos (you can even choose which one you trust) or git repos. Doing it this way the system is fully auditable, whereas when shipping a disk image it's harder to be confident that no nefarious extras have been added.
Even when using Freedombone metadata analysis by third parties is still possible. They might have a much harder time knowing what the content is, but they can potentially construct extensive dossiers based upon who communicated with your server when. Metadata leakage is a general problem with most current web systems and it is hoped that more secure technology will become available in future. But for now if metadata protection is your main concern using Freedombone won't help.
<h2id="unnumbered-3">How do I get a domain name?</h2>
<divclass="outline-text-2"id="text-unnumbered-3">
<p>
Suppose that you have bought a domain name (rather than using a free subdomain on freedns) and you want to use that instead.
</p>
<p>
Remove any existing nameservers for your domain (or select "custom" nameservers), then add:
</p>
<divclass="org-src-container">
<preclass="src src-bash">NS1.AFRAID.ORG
NS2.AFRAID.ORG
NS3.AFRAID.ORG
NS4.AFRAID.ORG
</pre>
</div>
<p>
It might take a few minutes for the above change to take effect. Within freedns click on "Domains" and add your domains (this might only be available to paid members). Make sure that they're marked as "private".
</p>
<p>
Select "Subdomains" from the menu on the left then select the MX entry for your domain and change the destination to <b>10:mydomainname</b> rather than <b>10:mail.mydomainname</b>.
</p>
<p>
To route email to one of your freedns domains:
</p>
<divclass="org-src-container">
<preclass="src src-bash">editor /etc/mailname
</pre>
</div>
<p>
Add any extra domains which you own, then save and exit.
<h2id="unnumbered-4">How do I get a "real" SSL certificate?</h2>
<divclass="outline-text-2"id="text-unnumbered-4">
<p>
You can obtain a free "official" (as in recognised by default by web browsers) SSL certificate from <ahref="https://www.startssl.com/">StartSSL</a>. You will first need to have bought a domain name, since it's not possible to obtain one for a freedns subdomain, so see <i>Using your own domain</i> for details of how to do that. You should also have tested that you can send email to the domain and receive it on the Freedombone (via Mutt or any other email client).
</p>
<p>
When creating a SSL certificate it's important that the private key (the private component of the public/private pair in <ahref="https://en.wikipedia.org/wiki/Public-key_cryptography">public key cryptography</a>) be generated on the Freedombone <i>and remain there</i>. Don't generate the private key via the StartSSL certificate wizard because this means that potentially they may retain a copy of it which could then be exfiltrated either via <ahref="https://en.wikipedia.org/wiki/Lavabit">Lavabit</a> style methodology, "implants", compromised sysadmins or other "side channel" methods. So that the private key isn't broadcast on the internet we can instead generate a certificate request, which is really just a request for authorisation of a public key.
</p>
<p>
Firstly you should have a web server site configuration ready to go. See <i>Setting up a web site</i> for details.
</p>
<p>
Within StartSSL under the validations wizard validate your domain, which means sending an email to it and confirming a code.
</p>
<p>
Now we can generate the certificate request as follows.
You can then click on "skip" within the StartSSL certificates wizard and copy and paste the encrypted request into the text entry box. A confirmation will be emailed back to you normally within a few hours.
</p>
<p>
Log into your StartSSL account and select <b>Retrieve Certificate</b> from the <b>Tool Box</b> tab. Copy the text.
Now visit your web site at <ahref="https://mydomainname.com/">https://mydomainname.com/</a> and you should notice that there is no certificate warning displayed. You will now be able to install systems which don't allow the use of self-signed certificates, such as <ahref="https://redmatrix.me/&JS=1">Red Matrix</a>.
Almost everywhere on the web you will read that self-signed certificates are worthless. They bring up scary looking browser warnings and gurus will advise you not to use them. Self-signed certificates are quite useful though. What the scary warnings mean - and it would be good if they explained this more clearly - is that you have an encrypted connection established but there is <i>no certainty about who that connection is with</i>. The usual solution to this is to get a "real" SSL certificate from one of the certificate authorities, but it's far from clear that such authorities can be trusted. There have been various scandals involving such organisations, and it does not seem plausible to assume that they are somehow immune to the sort of treatment which <ahref="http://en.wikipedia.org/wiki/Lavabit">Lavabit</a> received. So although most internet users have been trained to look for the lock icon as an indication that the connection is secured that belief may not always be well founded.
Security of web sites on the internet is still a somewhat unsolved problem, and what we have now is a less than ideal but <i>good enough to fool most of the people most of the time</i> kind of arrangement. Long term a better solution might be to have a number of certificate authorities in a number of different jurisdictions vote on whether a given certificate actually belongs to a given domain name. Experimental systems like this exist, but they're not widely used. Since the current certificate system has an enormous amount of inertia behind it change could be slow in arriving.
For now a self-signed certificate will probably in most cases protect your communications from "bulk" passive surveillance. Once you've got past the scary browser warning and accepted the certificate under most conditions (except when starting up the Tor browser) you should not repeatedly see that warning. If you do then someone may be trying to meddle with your connection to the server. You can also take a note of the fingerprint of the certificate and verify that if you are especially concerned. If the fingerprint remains the same then you're probably ok.
