12 Configuring SSL

SSL: An Introduction

Secure Sockets Layer (SSL) provides secure connections by allowing two applications connecting over a network to authenticate each other's identity and by encrypting the data exchanged between the applications. Authentication allows a server and optionally a client to verify the identity of the application on the other end of a network connection. Encryption makes data transmitted over the network intelligible only to the intended recipient.

SSL in WebLogic Server is an implementation of the SSL 3.0 and Transport Layer Security (TLS) 1.0 specifications.

WebLogic Server supports SSL on a dedicated listen port which defaults to 7002. To establish an SSL connection, a Web browser connects to WebLogic Server by supplying the SSL listen port and the HTTPs protocol in the connection URL, for example, https://myserver:7002.

Using SSL is compute intensive and adds overhead to a connection. Avoid using SSL in development environments when it is not necessary. However, always use SSL in a production environment.

One-Way and Two-Way SSL

SSL can be configured one-way or two-way:

• With one-way SSL, the server is must present a certificate to the client, but the client is not required to present a certificate to the server. The client must authenticate the server, but the server accepts a connection from any client. One-way SSL is common on the Internet where customers want to create secure connections before they share personal data. Often, clients will also use SSL to log on in order that the server can authenticate them.

• With two-way SSL (SSL with client authentication), the server presents a certificate to the client and the client presents a certificate to the server. WebLogic Server can be configured to require clients to submit valid and trusted certificates before completing the SSL connection.

Setting Up SSL: Main Steps

To set up SSL:

1. Obtain an identity (private key and digital certificates) and trust (certificates of trusted certificate authorities) for WebLogic Server. Use the digital certificates, private keys, and trusted CA certificates provided by the WebLogic Server kit, the CertGen utility, Sun Microsystem's keytool utility, or a reputable vendor such as Entrust or Verisign to perform this step.

   Note:

   If you use the CertGen utility to generate certificates, see Limitation on CertGen Usage, for information about limitations on its use. Certificates generated by CertGen are for demo purposes only and should not be used in a production environment.

2. Store the identity and trust. Private keys and trusted CA certificates which specify identity and trust are stored in a keystore.

   Note:

   This release of WebLogic Server supports private keys and trusted CA certificates stored in files, or in the WebLogic Keystore provider for the purpose of backward compatibility only.

3. Configure the identity and trust keystores for WebLogic Server in the WebLogic Server Administration Console. See "Configure keystores" in the Oracle Fusion Middleware Oracle WebLogic Server Administration Console Help.

4. Set SSL configuration options for the private key alias and password in the WebLogic Server Administration Console. Optionally, set configuration options that require the presentation of client certificates (for two-way SSL). See "Servers: Configuration: SSL" and "Configure two-way SSL" in the Oracle Fusion Middleware Oracle WebLogic Server Administration Console Help.

   Note:

   To enable a WebLogic Server instance to use a FIPS-compliant (FIPS 140-2) crypto module in the server's SSL implementation, make sure that the server start script (for example, startWebLogic.cmd/sh) contains the following:

   • jsafeFIPS.jar is added to the PRE_CLASSPATH variable.

   • The command line argument -Dweblogic.security.SSL.nojce=true is specified.

   FIPS 140-2 is a standard that describes U.S. Federal government requirements for sensitive, but unclassified use.

For information on configuring identity and trust for WebLogic Server, see Obtaining Private Keys, Digital Certificates, and Trusted Certificate Authorities and Storing Private Keys, Digital Certificates, and Trusted Certificate Authorities.

Using Host Name Verification

A host name verifier ensures the host name in the URL to which the client connects matches the host name in the digital certificate that the server sends back as part of the SSL connection. A host name verifier is useful when an SSL client (or a WebLogic Server acting as an SSL client) connects to an application server on a remote host. It helps to prevent man-in-the-middle attacks.

By default, WebLogic Server has host name verification enabled. As a function of the SSL handshake, WebLogic Server compares the common name in the SubjectDN in the SSL server's digital certificate with the host name of the SSL server used to accept the SSL connection. If these names do not match, the SSL connection is dropped. The SSL client is the actual party that drops the SSL connection if the names do not match.

If anything other than the default behavior is desired, either turn off host name verification or configure a custom host name verifier. Turning off host name verification leaves WebLogic Server vulnerable to man-in-the-middle attacks. Oracle recommends leaving host name verification on in production environments.

In this release of WebLogic Server, the host name verification feature is updated so that if the host name in the certificate matches the local machine's host name, host name verification passes if the URL specifies localhost, 127.0.01, or the default IP address of the local machine.

For more information about using host name verification, see the following topics in the Oracle Fusion Middleware Oracle WebLogic Server Administration Console Help:

• "Verify host name verification is enabled"

• "Disable host name verification"

• "Configure a custom host name verifier"

• "Servers: Configuration: SSL"

Enabling SSL Debugging

SSL debugging provides more detailed information about the SSL events that occurred during an SSL handshake. The SSL debug trace displays information about:

• Trusted certificate authorities

• SSL server configuration information

• Server identity (private key and digital certificate)

• The encryption strength that is allowed

• Enabled ciphers

• SSL records that were passed during the SSL handshake

• SSL failures detected by WebLogic Server (for example, trust and validity checks and the default host name verifier)

• I/O related information

Use the following command-line properties to enable SSL debugging:

-Dssl.debug=true -Dweblogic.StdoutDebugEnabled=true

You can include SSL debugging properties in the start script of the SSL server, the SSL client, and the Node Manager. For a Managed Server started by the Node Manager, specify this command-line argument on the Remote Start page for the Managed Server.

SSL debugging dumps a stack trace whenever an ALERT is created in the SSL process. The types and severity of the ALERTS are defined by the Transport Layer Security (TLS) specification.

The stack trace dumps information into the log file where the ALERT originated. Therefore, when tracking an SSL problem, you may need to enable debugging on both sides of the SSL connection (on both the SSL client or the SSL server). The log file contains detailed information about where the failure occurred. To determine where the ALERT occurred, confirm whether there is a trace message after the ALERT. An ALERT received after the trace message indicates the failure occurred on the peer. To determine the problem, you need to enable SSL debugging on the peer in the SSL connection.

When tracking an SSL problem, review the information in the log file to ensure:

• The correct config.xml file was loaded

• The setting for domestic, or export, is correct

• The trusted certificate authority was valid and correct for this server.

• The host name check was successful

• The certificate validation was successful

  Note:

  Sev 1 type 0 is a normal close ALERT, not a problem.

SSL Session Behavior

WebLogic Server allows SSL sessions to be cached. Those sessions live for the life of the server.

Clients that use SSL sockets directly can control the SSL session cache behavior. The SSL session cache is specific to each SSL context. All SSL sockets created by SSL socket factory instances returned by a particular SSL context can share the SSL sessions.

Clients default to resuming sessions at the same IP address and port. Multiple SSL sockets that use the same host and port share SSL sessions by default, assuming the SSL sockets are using the same underlying SSL context.

Clients that are not configured to use SSL sessions must call setEnableSessionCreation(false) on the SSL socket to ensure that no SSL sessions are cached. This setting only controls whether an SSL session is added to the cache; it does not stop an SSL socket from finding an SSL session that was already cached. For example, SSL socket 1 caches the session, SSL socket 2 sets setEnableSessionCreation to false but it can still reuse the SSL session from SSL socket 1 because that session was put in the cache.

SSL sessions exist for the lifetime of the SSL context; they are not controlled by the lifetime of the SSL socket. Therefore, creating a new SSL socket and connecting to the same host and port used by a previous session can resume a previous session as long as you create the SSL socket using an SSL socket factory from the SSL context that has the SSL session in its cache.

By default, clients that use HTTPS URLs get a new SSL session for each URL because each URL uses a different SSL context and therefore SSL sessions can not be shared or reused. You can retrieve the SSL session by using the weblogic.net.http.HttpsClient class or the weblogic.net.http.HttpsURLConnection class. Clients can also resume URLs by sharing a SSLSocket Factory between them.

Session caching is maintained by the SSL context, which can be shared by threads. A single thread has access to the entire session cache, not just one SSL session, so multiple SSL sessions can be used and shared in a single (or multiple) thread.

The following command-line arguments are ignored:

• weblogic.security.SSL.sessionCache.size

• weblogic.security.SSL.sessionCache.ttl

Configuring RMI over IIOP with SSL

Use SSL to protect Internet Interop-Orb-Protocol (IIOP) connections to Remote Method Invocation (RMI) remote objects. SSL secures connections through authentication and encrypts the data exchanged between objects.

To use SSL to protect RMI over IIOP connections:

1. Configure WebLogic Server to use SSL.

2. Configure the client Object Request Broker (ORB) to use SSL. Refer to the product documentation for your client ORB for information about configuring SSL.

3. Use the host2ior utility to print the WebLogic Server IOR to the console. The host2ior utility prints two versions of the interoperable object reference (IOR), one for SSL connections and one for non-SSL connections. The header of the IOR specifies whether or not the IOR can be used for SSL connections.

4. Use the SSL IOR when obtaining the initial reference to the CosNaming service that accesses the WebLogic Server JNDI tree.

For more information about using RMI over IIOP, see Oracle Fusion Middleware Programming RMI for Oracle WebLogic Server.

SSL Certificate Validation

WebLogic Server ensures that each certificate in a certificate chain was issued by a certificate authority. All X509 V3 CA certificates used with WebLogic Server must have the Basic Constraint extension defined as CA, thus ensuring that all certificates in a certificate chain were issued by a certificate authority. By default, any certificates for certificate authorities not meeting this criteria are rejected. This section describes the command-line argument that controls the level of certificate validation.

-Dweblogic.security.SSL.enforceConstraints=option 

-Dweblogic.security.SSL.enforceConstraints=strong

-Dweblogic.security.SSL.enforceConstraints=true

strict

-Dweblogic.security.SSL.enforceConstraints=strict

off

-Dweblogic.security.SSL.enforceConstraints=off

-Dweblogic.security.SSL.allowedcertificatepolicyids <identifier1>,<identifier2>,... 

java utils.ValidateCertChain -file pemcertificatefilename 
java utils.ValidateCertChain -pem pemcertificatefilename 
java utils.ValidateCertChain -pkcs12store pkcs12storefilename 
java utils.ValidateCertChain -pkcs12file pkcs12filename password 
java utils.ValidateCertChain -jks alias storefilename [storePass] 

java utils.ValidateCertChain -pem zippychain.pem 

Cert[0]: CN=zippy,OU=FOR TESTING
ONLY,O=MyOrganization,L=MyTown,ST=MyState,C=US

Cert[1]: CN=CertGenCAB,OU=FOR TESTING
ONLY,O=MyOrganization,L=MyTown,ST=MyState,C=US
Certificate chain appears valid

java utils.ValidateCertChain -jks mykey mykeystore

Cert[0]: CN=corba1,OU=FOR TESTING ONLY, O=MyOrganization,L=MyTown,ST=MyState,C=US
CA cert not marked with critical BasicConstraint indicating it is a CA
Cert[1]: CN=CACERT,OU=FOR TESTING ONLY, O=MyOrganization,L=MyTown,ST=MyState,C=US

Certificate chain is invalid

Using the nCipher JCE Provider with WebLogic Server

SSL is a key component in the protection of resources available in Web servers. However, heavy SSL traffic can cause bottlenecks that affect the performance of Web servers. JCE providers like nCipher that use a hardware card for encryption offload SSL processing from Web servers, which frees the servers to process more transactions. They also provide strong encryption and cryptographic processes to preserve the integrity and secrecy of keys.

WebLogic Server supports the use of the following JCE providers:

• The JDK JCE provider (SunJCE) in the JDK 5.0. For more information about the features in the JDK JCE provider, see http://java.sun.com/products/archive/jce/.

  By default, the JCE provider in the JDK 5.0 has export strength jurisdiction policy files. After filling out the appropriate forms, the domestic strength jurisdiction policy files are downloadable from Sun Microsystems at http://java.sun.com/javase/technologies/security/#UnlimitedDownload.

  WebLogic Server will continue to control the strength of the cryptography used by the WebLogic Server Application Programming Interfaces (APIs). Client code without the appropriate domestic strength cryptography setting will only be able to use the J2SE export strength default cryptography. On the server, WebLogic Server will enable either export or domestic strength cryptography.

• The nCipher JCE provider. See http://www.ncipher.com.

To install the nCipher JCE provider:

1. Install and configure the hardware for the nCipher JCE provider according to the product's documentation.

2. Install the files for the nCipher JCE provider. The following files are required:

   • Jurisdiction policy files—The JDK installs these files by default but they are of limited export strength.

   • Certificate that signed the JAR file

     Note:

     This step may have been performed as part of installing the hardware for nCipher JCE provider. In that case, verify that the files are correctly installed.

   • The JCE provider JAR files

   Choose an installation method for the files:

   • Install files as an extension. Copy the files to one of the following locations:

     JAVA_HOME/jre/lib/ext
     

     For example:

     MW_HOME/jdk150_03/jre/lib/ext
     

   • Install files in the CLASSPATH of the server.

3. Edit the Java security properties file (java.security) to add the nCipher JCE provider to the list of approved JCE providers for WebLogic Server. The Java security properties file is located in:

   JAVA_HOME/jre/lib/security/java.security
   

   Specify the nCipher JCE provider as:

   security.provider.n=com.ncipher.provider.km.mCipherKM
   

   where n specifies the preference order that determines the order in which providers are searched for requested algorithms when no specific provider is requested. The order is 1-based; 1 is the most preferred, followed by 2, and so on.

   The nCipher JCE provider must follow the RSA JCA provider in the security properties file. For example:

   security.provider.1=sun.security.provider.Sun
   security.provider.2=com.sun.rsajca.Provider
   security.provider.3=com.ncipher.provider.km.mCipherKM
   

4. Boot WebLogic Server.

5. To ensure the nCipher JCE provider is working properly, enable debugging according to the nCipher product documentation.

Specifying the Version of the SSL Protocol

WebLogic Server supports both the SSL V3.0 and TLS V1.0 protocols. When WebLogic Server is acting as an SSL server, the protocol that the client specifies as preferred in its client hello message. When WebLogic Server is acting as an SSL client, it specifies TLS1.0 as the preferred protocol in its SSL V2.0 client hello message, but can use SSL V3.0 as well, if that is the highest version that the SSL server on the other end supports. The peer must respond with an SSL V3.0 or TLS V1.0 message or the SSL connection is dropped.

While in most cases the SSL V3.0 protocol is acceptable some circumstances (compatibility, SSL performance, and environments with maximum security requirements) make the TLS V1.0 protocol more desirable. The weblogic.security.SSL.protocolVersion command-line argument lets you specify which protocol is used for SSL connections.

The following command-line argument can be specified so that WebLogic Server supports only SSL V3.0 or TLS V1.0 connections:

• -Dweblogic.security.SSL.protocolVersion=SSL3—Only SSL V3.0 messages are sent and accepted.

• -Dweblogic.security.SSL.protocolVersion=TLS1—Only TLS V1.0 messages are sent and accepted.

• -Dweblogic.security.SSL.protocolVersion=ALL—This is the default behavior.