Because the public key process is in widespread use, there are many public keys in circulation. Successful use of this system requires that every user be sure that a public key actually belongs to the assumed owner. The assignment of users to public keys is confirmed by trustworthy organizations with public key certificates. Such certificates contain the name of the key owner, the corresponding public key, and the electronic signature of the person issuing the certificate.

Trustworthy organizations that issue and sign public key certificates are usually part of a certification infrastructure that is also responsible for the other aspects of certificate management, such as publication, withdrawal, and renewal of certificates. An infrastructure of this kind is generally referred to as a public key infrastructure or PKI. One familiar PKI is the OpenPGP standard in which users publish their certificates themselves without central authorization points. These certificates become trustworthy when signed by other parties in the “web of trust.”

The X.509 Public Key Infrastructure (PKIX) is an alternative model defined by the IETF (Internet Engineering Task Force) that serves as a model for almost all publicly-used PKIs today. In this model, authentication is made by certificate authorities (CA) in a hierarchical tree structure. The root of the tree is the root CA, which certifies all sub-CAs. The lowest level of sub-CAs issue user certificates. The user certificates are trustworthy by certification that can be traced to the root CA.

The security of such a PKI depends on the trustworthiness of the CA certificates. To make certification practices clear to PKI customers, the PKI operator defines a certification practice statement (CPS) that defines the procedures for certificate management. This should ensure that the PKI only issues trustworthy certificates.

An X.509 certificate is a data structure with several fixed fields and, optionally, additional extensions. The fixed fields mainly contain the name of the key owner, the public key, and the data relating to the issuing CA (name and signature). For security reasons, a certificate should only have a limited period of validity, so a field is also provided for this date. The CA guarantees the validity of the certificate in the specified period. The CPS usually requires the PKI (the issuing CA) to create and distribute a new certificate before expiration.

The extensions can contain any additional information. An application is only required to be able to evaluate an extension if it is identified as critical. If an application does not recognize a critical extension, it must reject the certificate. Some extensions are only useful for a specific application, such as signature or encryption.

Table 16.1 shows the fields of a basic X.509 certificate in version 3.

If a certificate becomes untrustworthy before it has expired, it must be blocked immediately. This can become necessary if, for example, the private key has accidentally been made public. Blocking certificates is especially important if the private key belongs to a CA rather than a user certificate. In this case, all user certificates issued by the relevant CA must be blocked immediately. If a certificate is blocked, the PKI (the responsible CA) must make this information available to all those involved using a certificate revocation list (CRL).

These lists are supplied by the CA to public CRL distribution points (CDPs) at regular intervals. The CDP can optionally be named as an extension in the certificate, so a checker can fetch a current CRL for validation purposes. One way to do this is the online certificate status protocol (OCSP). The authenticity of the CRLs is ensured with the signature of the issuing CA. Table 16.2 shows the basic parts of a X.509 CRL.

The certificates and CRLs for a CA must be made publicly accessible using a repository. Because the signature protects the certificates and CRLs from being forged, the repository itself does not need to be secured in a special way. Instead, it tries to grant the simplest and fastest access possible. For this reason, certificates are often provided on an LDAP or HTTP server. Find explanations about LDAP in Chapter 4, LDAP—A Directory Service. Chapter 30, The Apache HTTP Server (↑Reference) contains information about the HTTP server.

YaST contains modules for the basic management of X.509 certificates. This mainly involves the creation of CAs, sub-CAs, and their certificates. The services of a PKI go far beyond simply creating and distributing certificates and CRLs. The operation of a PKI requires a well-conceived administrative infrastructure allowing continuous update of certificates and CRLs. This infrastructure is provided by commercial PKI products and can also be partly automated. YaST provides tools for creating and distributing CAs and certificates, but cannot currently offer this background infrastructure. To set up a small PKI, you can use the available YaST modules. However, you should use commercial products to set up an “official” or commercial PKI.

The first step when setting up a PKI is to create a root CA. Do the following:

In general, it is best not to allow user certificates to be issued by the root CA. It is better to create at least one sub-CA and create the user certificates from there. This has the advantage that the root CA can be kept isolated and secure, for example, on an isolated computer on secure premises. This makes it very difficult to attack the root CA.

If you need to change your password for your CA, proceed as follows:

A sub-CA is created in exactly the same way as a root CA.

The validity period for a sub-CA must be fully within the validity period of the “parent” CA. A sub-CA is always created after the “parent” CA, therefore, the default value leads to an error message. To avoid this, enter a permissible value for the period of validity.

Do the following:

Creating client and server certificates is very similar to creating CAs in Section 16.2.1, “Creating a Root CA”. The same principles apply here. In certificates intended for e-mail signature, the e-mail address of the sender (the private key owner) should be contained in the certificate to enable the e-mail program to assign the correct certificate. For certificate assignment during encryption, it is necessary for the e-mail address of the recipient (the public key owner) to be included in the certificate. In the case of server and client certificates, the hostname of the server must be entered in the Common Name field. The default validity period for certificates is 365 days.

To create client and server certificates, do the following:

To revoke compromised or otherwise unwanted certificates, do the following:

Revocation alone is not enough to deactivate a certificate. Also publish revoked certificates in a CRL. Section 16.2.6, “Creating CRLs” explains how to create CRLs. Revoked certificates can be completely removed after publication in a CRL with Delete.

The previous sections explained how to create sub-CAs, client certificates, and server certificates. Special settings are used in the extensions of the X.509 certificate. These settings have been given rational defaults for every certificate type and do not normally need to be changed. However, it may be that you have special requirements for these extensions. In this case, it may make sense to adjust the defaults. Otherwise, start from scratch every time you create a certificate.

All changes to the defaults only affect objects created after this point. Already-existing CAs and certificates remain unchanged.

If compromised or otherwise unwanted certificates need to be excluded from further use, they must first be revoked. The procedure for this is explained in Section 16.2.3, “Creating or Revoking a Sub-CA” (for sub-CAs) and Section 16.2.4, “Creating or Revoking User Certificates” (for user certificates). After this, a CRL must be created and published with this information.

The system maintains only one CRL for each CA. To create or update this CRL, do the following:

Applications that evaluate CRLs reject every certificate if the CRL is not available or has expired. As a PKI provider, it is your duty always to create and publish a new CRL before the current CRL expires (period of validity). YaST does not provide a function for automating this procedure.

The executing computer should be configured with the YaST LDAP client for LDAP export. This provides LDAP server information at runtime that can be used when completing dialog fields. Otherwise (although export may be possible), all LDAP data must be entered manually. You must always enter several passwords (see Table 16.3, “Passwords during LDAP Export”).

Certificates, CAs, and CRLs can be exported to LDAP.

If you have set up a repository on the computer for administering CAs, you can use this option to create the CA objects directly as a file at the correct location. Different output formats are available, such as PEM, DER, and PKCS12. In the case of PEM, it is also possible to choose whether a certificate should be exported with or without key and whether the key should be encrypted. In the case of PKCS12, it is also possible to export the certification path.

Export a file in the same way for certificates, CAs as with LDAP, described in Section 16.2.7, “Exporting CA Objects to LDAP”, except you should select Export as File instead of Export to LDAP. This then takes you to a dialog for selecting the required output format and entering the password and filename. The certificate is stored at the required location after clicking OK.

For CRLs click Export, select Export to file, choose the export format (PEM or DER) and enter the path. Proceed with OK to save it to the respective location.

You can select any storage location in the file system. This option can also be used to save CA objects on a transport medium, such as a USB stick. The /media directory generally holds any type of drive except the hard drive of your system.

If you have exported a server certificate with YaST to your media on an isolated CA management computer, you can import this certificate on a server as a common server certificate. Do this during installation or at a later point with YaST.

You need one of the PKCS12 formats to import your certificate successfully.

The general server certificate is stored in /etc/ssl/servercerts and can be used there by any CA-supported service. When this certificate expires, it can easily be replaced using the same mechanisms. To get things functioning with the replaced certificate, restart the participating services.

If you select Import here, you can select the source in the file system. This option can also be used to import certificates from a transport medium, such as a USB stick.

To import a common server certificate, do the following: