Documentation

This section answers some general questions and details the procedure to generate SSL keys.

To start serving HTTPS traffic from Cherokee you will have to enable SSL support and obtain a signed certificate (or signed your own). Ensure that the certificate files have the appropriate permissions for your OS and that you know the locations of the following:

All files should be generated using PEM-encoding.

Certificates are configured at the vServer level in the Security tab. This means you must have a vServer setup that corresponds to your Certificate’s security scope. Enter the full paths for your Certificate and corresponding Private Key file. To add a certificate chain file to you CA List set Client Certs. Request to Accept then put the full path to your chain file in the CA List box.

Yes. HTTP and HTTPS use different server ports. The former binds to port 80 and the latter to port 443, so there is no conflict between them. To provide HTTPS you will need matching certificates.

Keep in mind that to have one of your virtual servers with HTTPS enabled you will need to configure HTTPS settings for all of them.

HTTPS can run on any port, but the standards specify port 443. That’s where any HTTPS compliant browser will look by default. You can change that by specifying another port in the URL. For example, https://example.com:8080/ would look for an HTTPS server on port 8080.

Yes, it can. This is a common question because a web server cannot see the hostname header when the HTTPS request is being processed. The user entered host part of the URI must match the Common Name (CN) provided by the certificate. Since virtual hosts are in use, the CN of the first available certificate may or may not match the one specified in the early stages of TLS negotiation.

Fortunately this can be overcomed by using SNI (Server Name Indication), which places the host information in the SSL handshake. If the web server is accessed through a web browser that supports SNI, things will just work. Every modern web browser does.

For legacy scenarios where using SNI if not possible, there is a workaround detailed in the next question.

Using SNI is the clean and standard way to go. For legacy scenarios where using SNI is not and option, Cherokee supports a workaround. The mechanism requires the web server to listen to several IPs, and to assign a specific one for each virtual server. This can be done through the Match tab of the virtual server configuration, by assigning an IP-method to the desired IP. Remember that the list of virtual servers is evaluated from top to bottom, so in case you specify the same IP for several virtual servers, only the first one will actually match. In case you need more flexibility than that to match your domains, you can always add more than one matching criteria on that section. An IP/Subnet match plus a simultaneous wildcard match is a combination likely to cover every corner-case scenario you are presented with.

SSL support is not enabled by default. You will have to perform three straightforward actions is order to enable it:

This is a scenario known to happen only with some web browsers running specific Operating Systems. Specifically, there was a report where the server was using 2 domains, 2 IPs, and 2 SSL Certificates on one server. Everything worked as expected with Firefox, Opera, and Chromium, but IE/Chrome on Windows kept showing a certificate error on the domain that was not configured as default.

The workaround consists on creating a dummy default host with the second IP and certificate. The affected web browsers will then fall back to the IP and send the correct certificate.

HTTP can easily be tested like this:

For HTTPS it is not so easy because of the SSL protocol between TCP and HTTP. However you can do a similar check with the help of OpenSSL’s s_client command.

You will receive detailed information about the SSL handshake before the actual HTTP response.

A more general command line client is probably a better choice. cURL deals with both HTTP and HTTPS, and performs a bunch of other interesting stuff.

First some terminology:

Yes. Although in essence it is exactly the same, if you have a passphrase on your SSL private key file, a startup dialog will asks you to enter it. This can be problematic if the web server needs to be started automatically. In this case, the passphrase can be removed from your private key at the cost of erasing a security layer, which brings additional security risks.

Yes. A script is provided to assist you with Certificate Generation.

Just locate the contrib subdirectory and type:

And follow the instructions. It will generate the required files, but you will have to install them manually.

It has been tested and has worked fine every time, but if you don’t find the script or it doesn’t work for you can always follow the rest of the procedure described in this recipe to manually generate the certificates.

In any other case `find / -iname CA.pl -print` will help you to locate it.

It is another way to generate certificate files. Ramon Pons sent this little script to create self signed certificates::

You can see that, in essence, it issues the following command:

Which would produce a couple of files: the SSL Certificate File (server.crt) and the SSL Certificate key file (server.key).

This server.key does not have any passphrase. To add a passphrase to the key, you should run the following command, and enter & verify the passphrase as requested.

You should probably backup the key file and the entered passphrase in a secure location.

As noted above, if you have a pass-phrase on your SSL private key file, the web-server start up will remain on hold until you enter it. Here is the information needed to change it or even remove it, but bare in mind the security implications.

Simply read it with the old pass-phrase and write it again, specifying a new pass-phrase. This can be done withe these commands:

The RSA private key inside the server.key file is stored in encrypted format for security reasons. The pass-phrase is needed to decrypt this file, so it can be read and parsed. Thus, removing it removes a layer of security from the web server. It is advised to keep a backup copy of the original file before proceeding.

Since the server.key now contains an unencrypted copy of the key, if anyone gets it they will be able to impersonate you on the net.

To view the Certificate and the key run the commands:

The modulus and the public exponent portions in the key and the Certificate must match. It is difficult to visually check that the long modulus numbers are the same, so this approach can be used instead to obtain the numbers to compare (though it is mathematically less rigorous).

To check to which key or certificate a particular CSR belongs you can perform the same calculation on the CSR as follows:

Most services will require a fee to provide you with signed certificates, but some services will do this for free on their entry-product line.

You need to make sure that whoever signs your certificate is trusted as root CA on your operating system of choice. One such service that works by default on Windows, Linux, and Mac is StartSSL is one such service, although there are probably others out there.

If your public certificate is signed by a Certificate Authority’s intermediate certificate rather than a root certificate, you will need a chained CA. This is basically a list or path of certificates to instruct your visitors on how to follow the chain of trust. The list contains certificates that form the certificate chain (from the client/server certificate to the root CA certificate).

When using SSL, even a signed certificate can be worthless by itself. This is because the system will trust the root CA, but won’t normally know about the intermediate CA that actually signed your certificates.

Where to go next while following this chain of trust is specified in the chained certificate. The mechanism by which the certificate chain works is actually not that complex. When an unrecognized certificate is received, the certificate chain is used to obtain the root CA certificate. This chain is a list of certificates used to authenticate an entity. The path begins with the certificate of that entity, and each certificate in the chain is signed by the entity identified by the next certificate in the chain. The chain terminates with a root CA certificate, that is always signed by the CA itself. The signature of every certificate in the chain must be verified, all the way from the certificate owner until the root CA certificate is reached.

To create a certificate chain file, you simply need to your public certificate and your Certificate Authority’s intermediate certificate together. Assuming your public certificate is public.pem (the certificate that was signed by the Certificate Authority), and your CA’s public certificate (the intermediate Certificate supplied by your Certificate Authority) is inter_ca_public.pem, simply appending both will suffice to create a valid chained certificate.