All servers and clients which use the Websphere installation directory must be stopped for proper application of this eFix. Important note, if any of the efixes which this eFix supersedes has been applied, this eFix may take an exceptionally long time to install (possibly up to 15 minutes or more depending system speed). For PTF 4.0.2 only, PQ56667 must be applied before this fix is applied. Release: WebSphere 4.0.2 Operating System: All Supersedes eFixes: PQ61779, PQ62538, PQ62684, PQ63116, PQ63574, WAS_Security_09-13-2002_4.0.4-4.0.3_AEs_cumulative_eFix and WAS_Security_10-07-2002_4.0.4-4.0.3_AEs_cumulative_eFix and WAS_Security_10-31-2002_4.0.4-4.0.3_AEs_cumulative_eFix as well as any eFixes for APARs listed below. CMVC defect: See APAR list below Byte size of APAR: 1,388,990 Date: 11/24/2002 Description/symptom of problem: See APAR list below. Directions to apply efix: 1) Create temporary "efix" directory to store the jar file: AIX: /tmp/WebSphere/efix Solaris/Linux: /tmp/WebSphere/efix Windows: c:\temp\WebSphere\efix 2) Copy jar file to the directory 3) Shutdown WebSphere 4) Run the jar file with the following command answering questions/prompts as they appear: java -jar 5) Restart WebSphere 6) The temp directory may be removed but the jar file should be saved. Do not remove any files created and stored in the /WebSphere/AppServer/efix/ directories. These files are required if an efix is to be removed. Directions to remove an efix: NOTE: EFIXES MUST BE REMOVED IN THE ORDER THEY WERE APPLIED. DO NOT REMOVE AN EFIX UNLESS ALL EFIXES APPLIED AFTER IT HAVE FIRST BEEN REMOVED. YOU MAY REAPPLY ANY REMOVED EFIX. Example: If your system has efix1, efix2, and efix3 applied in that order and efix2 is to be removed, efix3 must be removed first, efix2 removed, and efix3 re-applied. 1) Change directory to the efix location (/WebSphere/AppServer/efix/). 2) Shutdown WebSphere 3) Run the backup jar file with the following command: java -jar 4) Restart WebSphere Directions to re-apply an efix: Follow the instructions for applying an efix. If the backup files still exist (from the previous efix application), you will be prompted to overwrite. Answer "yes" at the overwrite prompts. Additional Information: ------------------------------------------------------------------ Some systems may encounter an error similar to the following while applying this eFix. 2002/10/04 14:09:47 Command= /opt/WebSphere/AppServer/temp/FixSecurityJar.sh /opt/WebSphere/AppServer 2002/10/04 14:09:48 RC=xxx 2002/10/04 14:09:48 Error 112 -- The return code of xxx is not equal to the expected return code of 0. If you have not applied any of the 4 cumulative security efixes listed below on your system, then you can ignore these error messages in the extractor log. They are harmless and should not cause you any problems. PQ61779, PQ62538, PQ62684 or PQ63116 If you have applied any of the above 4 cumulative security efixes on your system and get these errors when applying a later cumulative security efix, then this would be a valid error. There are two resolutions to this error. The first is to remove this eFix and all previous eFixes in the opposite order they were installed until all of the eFixes listed above have been removed. Then all other eFixes including this one can be reapplied. The error above will still occur during installation but it can be ignored. The second is to follow the steps below. For Windows, execute the following: (Note use _AEs_ instead of _AE_ in commands below on AEs systems.) cd \WebSphere\AppServer bin\setupCmdLine.bat %JAVA_HOME%\bin\jar -xvf WAS_Security_10-31-2002_4.0.4-4.0.3-4.0.2_AE_cumulative_eFix.jar temp\FixSecurityJar.jar %JAVA_HOME%\bin\jar -xvf WAS_Security_10-31-2002_4.0.4-4.0.3-4.0.2_AE_cumulative_eFix.jar temp\FixSecurityJar.bat temp\FixSecurityJar.bat %WAS_HOME% For UNIX systems, execute the following: (Note use _AEs_ instead of _AE_ in commands below on AEs systems.) . bin/setupCmdLine.sh $JAVA_HOME/bin/jar -xvf eFix/WAS_Security_10-31-2002_4.0.4-4.0.3-4.0.2_AE_cumulative_eFix.jar temp/FixSecurityJar.jar $JAVA_HOME/bin/jar -xvf eFix/WAS_Security_10-31-2002_4.0.4-4.0.3-4.0.2_AE_cumulative_eFix.jar temp/FixSecurityJar.sh . temp/FixSecurityJar.sh $WAS_HOME The following properties must be set for PQ56218 to be effective: In the admin.config file: com.ibm.CORBA.ListenerPort=2000 com.ibm.CORBA.SSLPort=2001 com.ibm.CORBA.LSDPort=9000 com.ibm.CORBA.LSDSSLPort=9001 In each of the Application servers system properties: com.ibm.CORBA.ListenerPort=3000 com.ibm.CORBA.SSLPort=3001 Note, each port number must be unique and must not be in use by any other service on the system. To enable PQ63020, set the following property in the admin.config. com.ibm.ejs.security.customregistry.useSecurityName=true Included APARS: APAR: PQ63457 Version: 400 Abstract: ALLAUTHENTICATEDUSERS NOT WORKING FOR ROLE MAPPING Error Description: Environment: WebSphere Application Server (WAS) 4.0.x . Description: Mapping an application role to the AllAuthenticatedUsers role available by checkbox in the security center did not function as prescribed...users were denied access to pages after authenticated succeeded. Local Fix: None Logically Dependent Apars: Users Affected: All WebSphere Application Server users who have security enabled and are using WebSphere Studio Application Developer (WSAD) to create and deploy EARs with security role references. Problem Description: Users given access to a resource via the special role "All Authenticated Users" receive authorization failures. Recommendation: Problem Summary: Users given access to a resource via the special role "All Authenticated Users" receive authorization failures. The reason for this was that WAS erroneously creating empty access ID for special authorization subjects "All Authenticated Users" and "Everyone." These special authorization subjects are not supposed to have access IDs which cause the name to be used as the access ID. Problem Conclusion: Code was added at runtime to recreate the access ID as the name if the supplied ID is empty. Test Comments: Circumvention: Edit the EAR with the AAT and remove the reference to "All Authenticated Users" and "Everyone," then save. Edit it again and re-add the references. Temporary Fix: Contained in security cumulative eFix PQ63457 or more recent. Comments: APAR: PQ63574 Version: 400 Abstract: INTERMITTENT AUTHORIZATION FAILURES FOR AUTHENTICATED USERS THAT BELONG TO A ROLE Error Description: Intermittent authorization failures occur for authenticated users who are properly configured for a security role. The customer has discovered that the problem may not be intermittent, and may instead occur on the first web application loaded when starting an application server. The problem seems to occur because for the first web app loaded the user registry value is null on the call to addAuthorizationTable(), so fillMissingAccessIds() does not occur. This results in the roles information not being available when performing authorizations for those roles. . This problem may be the same as the one reported in internal defect 139504. Local Fix: The customer has worked around the problem by building a workaround into the WSAccessManager. They inserted a fix into the constructor whose signature is "public WSAccessManager(RegistryImpl registryimpl)". The fix simply inserts a call to fillAccessIds() to force the authorization information to be loaded. Logically Dependent Apars: Users Affected: All WebSphere Application Server users who have enabled security. Problem Description: Intermitant authorization problems due to improper authorization table initialization. Recommendation: Problem Summary: Intermittent authorization problems due to improper authorization table initialization. The tables were not initialized properly due to the improper use of a boolean value which was used to indicate initialization state of the tables. Problem Conclusion: The boolean value was used at times to indicate a single table had been initialized when it should only be used to indicate all tables had been initialized. This was corrected. Test Comments: Circumvention: Temporary Fix: Contained in security cumulative eFix PQ63457 or more recent. Comments: APAR: PQ66162 Version: 400 Abstract: SERVERSIDEAUTHENTICATOR DOESN'T THROW EXCEPTION Error Description: SSOAuthenticator - When used 1. Authenticates userid and password. 2. Throws exception when authentication fails (works correctly) 3. Set HttpRequest and HttpResponse LPTA cookie so that they can be passed by servlet. 4. DOES NOT SET THE CONTEXT for SAS communication for ejb layer. So the ejb thinks the user is UNAUTHENTICATED and fails. . ServerSideAuthenticator - When used 1. Authenticates userid and password. 2. DOES NOT Throws exception when authentication fails. 3. DOES NOT Set HttpRequest and HttpResponse LPTA cookie so that they can be passed by servlet. 4. Set context for for SAS communication for ejb layer. . So, with this being the case, i have to use 2 separate API to authenticate correctly and set the desired information to enable J2EE security framework. Right now I call ServerSideAuthenticator first then SSOAuthenticator. Seems kind of expensive to me and confusing. . Customer requests a fix for ServerSideAuthenticator for WAS 4.03 on AIX. . When ServerSideAuthentication fails to authenticate, it returns a null credential. Not basic credentials. Second, a client may use ServerSideAuthenitcate for authentication purpose only. They may never go to a secure resource (like ejb) after that. Or my ejb may not be secure....I know that the ejb container will through the error because the user is UNAUTHENTICATED. This is not new to us developers, however, WAS is relying on the Ejb Server Container (security mechanism) to throw the error for simple WAS authentication...Not authorization...this is not correct. Local Fix: none Logically Dependent Apars: Users Affected: WebSphere Application Server users who uses ServerSideAuthenticator to perform authentication. Problem Description: ServerSideAuthenticator should throw LoginFailed when login fails. Recommendation: Problem Summary: ServerSideAuthenticator should throw org.omg.SecurityLevel2.LoginFailed exception when the login fails and the force_authn flag is true instead of returing a null credential. Problem Conclusion: ServerSideAuthenticator will now throw org.omg.SecurityLevel2.LoginFailed exception when login fails. Test Comments: Circumvention: Temporary Fix: Available Comments: APAR: PQ57307 Version: 400 Abstract: ISUSERINROLE ALWAYS RETURNS FALSE FOR JSP Error Description: Users attempting to use the isUserInRole method of the HTTPServletRequest object to do authentication are unable to do so in JSP code because the return value is always false. Local Fix: NONE Logically Dependent Apars: Users Affected: WebSphere Application Server users with security enabled, calling isUserInRole from a JSP. Problem Description: Calling isUserInRole() within JSP always return false. Recommendation: Apply efix, or use Servlet instead of jsp to call isUserInRole(). Problem Summary: Calling isUserInRole() with jsp always return false. Problem Conclusion: The problem only happen in jsp(not servlet), and has been corrected. Test Comments: Circumvention: Temporary Fix: send testing efix to customer Comments: APAR: PQ57615 Version: 400 Abstract: ADDING USERS TO ROLES THROUGH BROWSER CONSOLE DOESN'T WORK. Error Description: AEs 4.0.1 Win2000 . When trying to associate Roles with Users or Groups throught the AEs/d Administrative console (browser) the update appears successful but actually fails. Not only does it not create the proper accessId for the user but it also removes the ones which existed before. . Customers ibm-application-bnd.xmi has following entry . This should look like: . This results in an "unauthorization exception" when a user tries to access a web resource where they should have Local Fix: Manually add accessId to xmi file. Logically Dependent Apars: Users Affected: WebSphere Application Server users of enabled security in AEs. Problem Description: Security accessID is not being filled in properly by some tools. Recommendation: apply Efix. Problem Summary: Some assemble / deployment tools do not fill security accessID properly(empty string). It should either fill fully qualified ID, or do nothing. Security runtime will fix the problem by treating empty ID as not ID presented. Problem Conclusion: Security runtime will fix the problem by treating empty access ID as not ID presented, thus will invoke registry to refill the access ID. Test Comments: Circumvention: Temporary Fix: send efix to customer Comments: APAR: PQ59684 Version: 400 Abstract: WITH SECURITY ON, AND PASSING THE SECURITY CHALLENGE, THE RETURN URL IS NOT CORRECT WHEN THE SERVLET IS IN A SUBDIRECTORY Error Description: The customer requests the following URL: http://mburati02:9080/bowstreet5/webengine/factory/admin/Factory Admin After successfully logging in, it incorrectly redirects to the following mangled URL: http://mburati02:9080/bowstreet5/webengine/factory/admin/Factory Admin/factory/admin/FactoryAdmin Local Fix: Make sure that the requested servlet is accessible from the context root of the webapp. Logically Dependent Apars: Users Affected: WebSphere Application Server developers using Form Login based security. Problem Description: Form login redirection fails when security is enabled. Recommendation: Problem Summary: When Form Login based security is enabled, the originally requested url is not properly sent to the client if the form login page is located in a directory different than the originally requested url. Problem Conclusion: Modified the storing of the originally requested url to properly handle whether the redirect should be relative to the context root of the web module or the web server. Test Comments: Circumvention: Temporary Fix: //wasdoc0/apars/pq59684/4.0.2_4.0.3 Comments: APAR: PQ60895 Version: 400 Abstract: WASSEC - AUTHENTICATION INFORMATION NOT RETURNED TO THE BROWSER IF A PROTECTED URI IS NOT ACCESSED BEFORE J_SECURITY_CHECK POST. Error Description: If authentication information is posted to j_security_check before a protected URI is accessed, the browser does not receive authentication information and therefore subsequent requests of protected resources fail. Failure stack in app svr stdout (excerpt): ExtendedMessage: Servlet Error: : java.lang.NullPointerException at com.ibm.servlet.personalization.sessiontracking.SessionContext.i sProtoco lSwitch(SessionContext.java:1884) at com.ibm.servlet.personalization.sessiontracking.SessionContext.s houldEnc odeURL(SessionContext.java:1834) at com.ibm.servlet.engine.srt.SRTSessionAPISupport.encodeURL(SRTSes sionAPIS upport.java:137) at com.ibm.servlet.engine.srt.SRTServletResponse.encodeURL(SRTServl etRespon se.java:220) at com.ibm.servlet.engine.webapp.HttpServletResponseProxy.encodeURL (HttpSer vletResponseProxy.java:51) at com.ibm.ws.security.web.FormLoginServlet.formLogin(FormLoginServ let.java :388) at com.ibm.ws.security.web.FormLoginServlet.doPost(FormLoginServlet .java:16 0) at javax.servlet.http.HttpServlet.service(HttpServlet.java:760) at javax.servlet.http.HttpServlet.service(HttpServlet.java:853) at com.ibm.servlet.engine.webapp.StrictServletInstance.doService(Se rvletMan ager.java:827) Local Fix: none Logically Dependent Apars: Users Affected: All WebSphere Application Server users using Form Login. Problem Description: The browser failed to receive authentication information when the Form Login page was directly accessed. Recommendation: Problem Summary: The browser failed to receive authentication information when the Form Login page was directly accessed instead of being redirected to the page by attempting to access a protected URI. Problem Conclusion: Test Comments: Circumvention: Temporary Fix: Test fix supplied on 5/22/2002. Comments: APAR: PQ62260 Version: 400 Abstract: OUT OF MEMORY CONDITION WHEN WAS R403 IS UNDER HEAVY LOAD WITH SECURITY ENABLED Error Description: Customer using a CustomRegistry and running a load test with 100 stress clients that do the following: - authenticate using a Trust Association & the custom registry - fetch a single JSP page - pause 10 seconds - leave the site. NO LOGOUT - repeat The clients use userids out of a possible population of about 10,000 userids. Thus, over a reasonable period of time, WAS will see 10,000 unique userids. The admin server, over a long period of time, logs out of memory errors. In this case, it took 4 hours. Local Fix: none. Logically Dependent Apars: Users Affected: All WebSphere Application Server users who have enabled security using LTPA authentication. Problem Description: Slow performance and high CPU usage during security cache cleaning cycles and possible OutOfMemoryExceptions. Recommendation: Problem Summary: Slow performance and high CPU usage during security cache cleaning cycles and possible OutOfMemoryExceptions. These issues are seen on servers with a large number of different users (12000 on the reported system) accessing the system within the security cache timeout period. The performance problem was a result of the cache cleaning algorithm which evaluated each cache entry as to whether or not it needed to remain in the cache. The out of memory condition was a result of the algorithms failure to remove cache entries. Problem Conclusion: The cache cleaning algorithm was changed to evaluate all cache entries inserted into the cache within a similar time period (one half the security cache timeout) at the same time. The new algorithm guarantees that the vast majority of entries will stay in the cache for the minimum of the security cache time out. There is now the potential for a small number of cache entries to be flushed from the cache early or for a cache miss to occur when there should have been a hit. This is negligible in comparison to the amount of time saved in cache cleaning, however. Test Comments: Circumvention: Temporary Fix: Comments: APAR: PQ62684 Version: 400 Abstract: CREATE ACCESSID AT RUNTIME IF THEY HAVE NOT BEEN CREATED. Error Description: Customer is adding to Role Group Roles using WSCP, WSCP did not create and AccessID in the process. The results are authorization exceptions when using the group. Local Fix: Manually add Roles to Group Roles through the AAT tool but this is too manual a process due to the large volume of users and nodes etc. Logically Dependent Apars: Users Affected: All WebSphere Application Server users adding groups to Roles via WSCP. Problem Description: Authorization failures for members of groups that were added to a Role via WSCP. Recommendation: Problem Summary: Authorization failures for members of groups that were added to a Role via WSCP. The problem stemmed from the fact that WSCP was not inserting an access ID for the group, which is used internally by WebSphere for authorization, into the Role. Problem Conclusion: The security runtime now creates an access ID if it is missing. Test Comments: Circumvention: Use the AAT or the Administration Console. Temporary Fix: testing eFix was send to customer. Comments: APAR: PQ63020 Version: 400 Abstract: ADD FUNCTIONALITY TO TOGGLE GETREMOTEUSER() TO RETURN ACTUAL USER ID INSTEAD OF THE DISPLAY NAME WHEN USING CUSTOM REGISTRY Error Description: Environment: WebSphere Application Server 4.0.3 AE . Description: Customer is implementing LTPA custom user registry. When they perform getRemoteUser() or getPrincipalName() mathods, they get the display name (as a result of getUserDisplayName() method call) instead of the actual user ID, and there seems to be no alternative to acquire the user ID when the getUserDisplayName() is implemented to return other than the user ID. The developer suggests creating a configuration item to toggle getRemoteUser() to return the user ID instead of the display name. Local Fix: None Logically Dependent Apars: Users Affected: WebSphere Application Server users who use custom registry authentication. Problem Description: Wehn using a custom registry, the API getRemoteUser() returns display name rather than security name. Recommendation: Apply this eFix, or use the display name as the same as the security name. Problem Summary: When using a custom registry implementation, security returned the user's display name for getRemoteUser() and getPrincipalName() instead of the user's security name. Problem Conclusion: If getRemoteUser() and getPrincipalName() are required to return the user's security name, a system property has been added to enable this. com.ibm.ejs.security.customregistry.useSecurityName=true Test Comments: Circumvention: Temporary Fix: test eFix provided. Comments: APAR: PQ68148 Version: 400 APAR: PQ65884 Version: 350 Abstract: REQUESTDISPATCH.FORWARD() TO A PROTECTED SERVLET FAILS Error Description: . Failing on security when doing requestdispatch.forward() to a protected servlet (and now failing) from the "baseLogon" servet that calls SSOAuthenticator. After calling SSOAuthenticator, the request thread should have a security context established and should not fail on the requestdispatch.forward() call. Local Fix: Test efix PMR81595-356-test-0829 fixed the customer's problem. Need official efix. Logically Dependent Apars: Users Affected: All WebSphere Application Server users who have enabled security. Problem Description: Users may not be properly challenged while accessing secured resources. Recommendation: Problem Summary: Authenticated user may get challenged again, or unauthenticated user may not be challenged as authentication was not properly flaged. This scenario is only likely to occur if a servlet forwards or dispatches to another secured servlet. Problem Conclusion: The flag used to determine authentication was not used correctly. The flag has now been removed as it is redundant. Test Comments: Circumvention: Temporary Fix: test eFix has been send to customer Comments: APAR: PQ66190 Version: 400 Abstract: PROPFILEPASSWORDENCODER COMMAND CAUSES PROPERTY FILES' OWNERSHIP AND PERMISSIONS TO CHANGE TO THAT OF THE USER RUNNING THE SCRIPT Error Description: Environment: WebSphere Application Server 4.0.3 AE for Solaris . Description: Running the PropFilePasswordEncoder.sh command causes the user/group ownership and permissions to change to the default settings for the user running the script. Local Fix: Change ownership and permissions back after running the command. Logically Dependent Apars: Users Affected: All WebSphere Application Server users of PropFilePasswordEncoder script Problem Description: PropFilePasswordEncoder command causes property files' ownership to change Recommendation: Problem Summary: After running PropFilePasswordEncoder script, the user and group ownership and the file permissions of those files changed to the default settings for the user running the PropFilePasswordEncoder script. Problem Conclusion: The way of how backup files are generated is modified, so the original property files can keep their original file permissions. Test Comments: Circumvention: Temporary Fix: Available Comments: APAR: PQ66974 Version: 400 Abstract: GETCALLERPRINCIPAL() SHOULD JUST ALWAYS RETURN "UNAUTHENTICATED" Error Description: Customer sent the following: We have a distributed log viewer servlet architecture, which means each AppServer must have a Servlet to read log file and present it to the end user. . So we must know how to invoke this servlet before it is serviceable(choosable from a drop down list). In the servlet init() method, we probe WebSphere servlet engine configuration to obtain HTTP port number for the hosting AppServer. Then the user can pick a LogViewer and invoke it based on this information. . The available LogViewer Tracking EJB has to be invoked from the init() method in this sense, so in the serlvet init() we will perform following execution: . try { /** retrieve the collection of application server adapters, retrieve configuration information from servlet engin */ String adapterName = SystemProperties.get("LOGSERVER_ADAPTER"); LogServerAdapter adapter = . (LogServerAdapter)Class.forName(adapterName).newInstance(); . LogServer myLogServer = adapter.getLocalLogServer(); m_logServerManager = CoreManagerFactory.getLogServerManager(); /** first, see if the log server is already in the database */ Collection logServers = m_logServerManager.getLogServers(); for (Iterator it = logServers.iterator(); it.hasNext(); ) { LogServer logServer = (LogServer)it.next(); if (myLogServer.getHost().equalsIgnoreCase(logServer.getHost()) && myLogServer.getPort() == logServer.getPort()) { m_logServerID = logServer.getID(); break; } } if (m_logServerID == null) { /** did not find my log server in the database, create a new record */ m_logServerID = m_logServerManager.createLogServer(myLogServer); } } catch (Exception e) { m_cat.l7dError(CoreEjbBundleKey.DEFAULT, e); } . And in our LogServerManager EJB, it is plain DB persistence stateless session bean. createLogServer() is just an insert statement, while getLogServerManager() is simple select statement. As we have an abstract OR layer, I just provide the functionality of these methods here. . In our EJB, we have logging system, which will always capture user's caller information at the begining of the method and push it down to the underneath logging system. So whenever there is exception and it needs logging, the logging system can always write down the caller infomation transparently. Logging system is based on Log4j. . createLogServer(Object value) { // this method internally will keep track caller info, it is pushed down to the logging system, which is defined in a parent EJB class methodStart("createLogServer", value); try { // operations; } catch(Exception e) { // whenever we want to log the exception, we retrieve back the caller info directly from logging system // so the api is ignorant of it is inside EJB or Servlet logging(e); } finally { // we will pop up the caller info from the logging system methodFinish(); } } . This model is applied to every single EJB in our system, and it will fail in the methodStart() invokation for requests initiated from init() of a servlet. . private void methodStart(String methodName, Object params) { m_strCurrentMethod = methodName; m_params = params; . if (m_sessionContext != null) NDC.push(m_sessionContext); . Category cat = getCategory(); if (cat != null && cat.isDebugEnabled()) { StringBuffer buf = new StringBuffer(100); buf.append("ENTER: "); buf.append(getMethodNameWithParams()); cat.debug(new String(buf)); } } . In the NDC.push(): public static void push( javax.ejb.EJBContext theContext { push( theContext.getCallerPrincipal() ); } . NDC class is used for: The NDC class implements nested diagnostic contexts as defined by Neil Harrison in the article "Patterns for Loggin Diagnostic Messages" part of the book "Pattern Languages of Program Design 3" edited by Martin et al.

A Nested Diagnostic Context, or NDC in short, is an instrument to distinguish interleaved log output from different sources. . This problem applies to all EJBs that will track caller and being invoked from init() method of servlets. Local Fix: None Logically Dependent Apars: Users Affected: WebSphere Application Server 4.0 security users who use programmatic security. Problem Description: If the anonymous principal is propagated or the user identity is missing, getCallerPrincipal() returns inconsistant results. Recommendation: Problem Summary: getCallerPrincipal() returns different results in different timing if the anonymous principal is propagated or caller identity is missing. EJB 1.2 spec define that getCallerPrincipal() should always return container specific unauthenticated principal. Problem Conclusion: getCallerPrincipal() should return "UNAUTHENTICATED", the WebSphere specific unauthenticated principal, if caller identity is missing or the anonymous principal is propagated. Test Comments: Circumvention: Temporary Fix: send testing eFix to customer. Comments: APAR: PQ67473 Version: 400 Abstract: % SIGN IN THE USER CREDENTIALS IS CORRUPTED WHEN PASSED THROUGHTHE GETUSERPRINCIPAL CLASS Error Description: custom tion Description Form based security using er uses is a \242 (cent sign). When passed to getUserPrincipal, .ustomer programmatically checks security with user name goes in as usercompany, and comes back from the .etUserPrincipal API. For business reasons, customer uses a method as usercompany. This worked under .egistry to store usernames as weblogic :) and they want it to work the same under websphere. Problem Summary The security check passes OK, but the delim the Local Fix: None Logically Dependent Apars: Users Affected: WebSphere Application Server users with security enabled. Problem Description: Unicode character in the principal name gets changed after login with SSOAuthenticator. Recommendation: Problem Summary: When using unicode characters in the principal name, the principal name gets changed with an extra character in front of the unicode character when getUserPrincipal().getName() is called after login. Problem Conclusion: Inconsistent encoding was used when we convert bytecode to string and back. UTF8 should be used. Test Comments: Circumvention: Temporary Fix: available Comments: APAR: PQ56218 Version: 400 Abstract: PMI DOES NOT RECOVER WHEN WEBSPHERE APP SERVER IS RESTARTED WHEN SECURITY ENABLED Error Description: ** Brief Description: Customer has a java PMI client (Performance Monitoring Infrastructure)programming interface provided by IBM. -- It logs into WAS fine the first time but when you stop and restart WAS with security on it does not log back in. Getting error Cannot create PmiService bean obj; nested exception is: java.rmi.RemoteException: CORBA INTERNAL 0 No Local Fix: none Logically Dependent Apars: Users Affected: All WebSphere Application Server users which have enabled security. Problem Description: Java clients fail to log into WebSphere Application Server after the server is restarted with security enabled. Recommendation: Problem Summary: After the Java client logged into the WebSphere Application Server, if the server restarts for any reason, then the Java Client cannot login again with security enabled. The client will get the java.rmi.RemoteException: CORBA INTERNAL 0 Problem Conclusion: When the client receives a request reject message, the client now checks the credential associated with the request. If the credential is a dummy credential, then the client tries to re-establish the secure association with the server again. Test Comments: Circumvention: Temporary Fix: PQ56218_eFix_AEServer.jar Comments: APAR: PQ58764 Version: 400 Abstract: CORBA EXCEPTION ERROR Error Description: The customer stated when java client his the application server he is getting the following error into the client stdout file: CORBA exception errors... failed mutual authentication handshake... session does not exist in the session table. He also noticed jsa150e-unable to find session in the session table errors. Def ect 120428 Local Fix: No fix or workaround available yet. Logically Dependent Apars: Users Affected: All WebSphere Application Server users which have enabled security. Problem Description: CORBA exception errors with failed mutual authentication message Recommendation: Problem Summary: Client gets CORBA exception errors, failed mutual authentication handshake after 45 minutes of attempting to complete the request. The following CORBA exception occurs: "org.omg.CORBA.NO_PERMISSION: Failed mutual authentication handshake. Session does not exist in the session table." Problem Conclusion: The server now passes back the security context along with the exception, so the client can react accordingly. Test Comments: Circumvention: Temporary Fix: PQ58764_401_test.jar Comments: APAR: PQ55325 Version: 350 Abstract: AUTHORIZATION FAILURES IN WAS 3.5.5 Error Description: After upgrading from WAS 3.5.4 to 3.5.5, randomly some users are able to authenticate to use their application, while others are not. . The app server stderr file shows the following exceptions: "UnauthorizedSessionRequestException: SessionContext: a user authenticated as anonymous has attempted to access a session owned by user:xxx" "RuntimeException: SessionContext: an attempt to access a session while WebSphere Session Manager was turned off was made" . Also, a SAS log file is generated which contains the following info: "1> 2001-11-27 14:14:19.171 , ServerID: 2106094737 , PrincipalAuthenticatorImpl.authenticate : Invalid or null client security name, unable to authenticate. 1> 2001-11-27 15:58:43.296 , ServerID: 2106094737 , CredentialsImpl.run : The expiration time for ltpa credentials is too short relative to the ORB request timeout and/or the security cache timeout; a method request could take longer than the period over which the credentials will remain valid, or the credentials could expire while in the server cache." Local Fix: None. Logically Dependent Apars: Users Affected: All WebSphere Application Server users who have enabled security and are on 3.5.5, 4.02 or a more recent PTF will potentially be affected by this. Problem Description: NegativeArraySizeException in SecurityAttributeList. getAttributeStringArray() which causes an authorization failure. Recommendation: Problem Summary: The exception included below causes an authorization failure. java.lang.NegativeArraySizeException at com.ibm.ISecurityUtilityImpl.SecurityAttributeList. getAttributeStringArray(SecurityAttributeList.java (Compiled Code)) at com.ibm.ejs.security.web.WebCollaborator. getGrantedPermissions(WebCollaborator.java (Compiled Code)) at com.ibm.ejs.security.web.WebCollaborator. checkAuthorization(WebCollaborator.java (Compiled Code)) at com.ibm.ejs.security.web.WebCollaborator. authorize(WebCollaborator.java(Compiled Code)) at com.ibm.ejs.security.web.EJSWebCollaborator. preInvoke(EJSWebCollaborator.java(Compiled Code)) at com.ibm.servlet.engine.webapp. WebAppSecurityCollaborator. preInvoke(WebAppSecurityCollaborator.java (Compiled Code)) at com.ibm.servlet.engine.webapp.WebAppRequestDispatcher. dispatch(WebAppRequestDispatcher.java(Compiled Code)) at com.ibm.servlet.engine.webapp.WebAppRequestDispatcher. forward(WebAppRequestDispatcher.java(Compiled Code)) at com.ibm.servlet.engine.srt.WebAppInvoker. doForward(WebAppInvoker.java(Compiled Code)) at com.ibm.servlet.engine.srt.WebAppInvoker. handleInvocationHook(WebAppInvoker. java(Compiled Code)) at com.ibm.servlet.engine.invocation.CachedInvocation. handleInvocation(CachedInvocation.java(Compiled Code)) at com.ibm.servlet.engine.invocation. CacheableInvocationContext. invoke(CacheableInvocationContext.java(Compiled Code)) at com.ibm.servlet.engine.srp.ServletRequestProcessor. dispatchByURI(ServletRequestProcessor.java (Compiled Code)) at com.ibm.servlet.engine.oselistener. OSEListenerDispatcher.service (OSEListener.java(Compiled Code)) at com.ibm.servlet.engine.oselistener.SQEventListenerImp$ ServiceRunnable.run(SQEventListenerImp.java (Compiled Code)) at com.ibm.servlet.engine.oselistener.SQEventListenerImp. notifySQEvent(SQEventListenerImp.java(Compiled Code)) at com.ibm.servlet.engine.oselistener.serverqueue. SQEventSource.notifyEvent(SQEventSource.java (Compiled Code)) at com.ibm.servlet.engine.oselistener.serverqueue. SQWrapperEventSource$SelectRunnable.notifyService (SQWrapperEventSource.java(Compiled Code)) at com.ibm.servlet.engine.oselistener.serverqueue. SQWrapperEventSource$SelectRunnable.run (SQWrapperEventSource.java:221) at com.ibm.servlet.engine.oselistener.outofproc. OutOfProcThread$CtlRunnable.run (OutOfProcThread.java:248) at java.lang.Thread.run(Thread.java:481) Problem Conclusion: Test Comments: Circumvention: Temporary Fix: Comments: APAR: PQ55948 Version: 350 Abstract: OUTOFMEMORY (MEMORY LEAK) OCCURS WHEN SECURITY IS ENABLED AND APPLICATION SPAWNS A ASYNCHRONOUS THREAD WHICH MAKES AN ORB CALL Error Description: Environment: WebSphere Application Server (WAS) 3.5.3 Advanced Edition . Description: Customer has a servlet which spawns a thread which in turn makes a call to MQSeries. In this scenario, when WAS security is enabled, a memory leak occurs which eventually results in OutOfMemory as objects instantiated by the servlet stay around because security references to the objects don't go away. Local Fix: None Logically Dependent Apars: Users Affected: All WebSphere Application Server users which have deployed servlets which are EJB clients and instantiate new threads and WebSphere security is enabled. Problem Description: Security associates thread specific data with the hashtables which prevent the threads from being garbage collected. Recommendation: Problem Summary: Security associates thread specific data with the hashtables using the thread object as a key. Since there is a reference to the thread object stored in a hashtable, the thread object will not be garbage collected even if it has gone out of context in the servlet. This issue can be exacerbated by subclassing the thread in which the subclassed thread has references to large data objects. Problem Conclusion: Logic was added to the hashtables to periodically remove all references to threads which were no longer executing. The logic is activated if N new threads are added to the hashtable since the last time the removal logic has run. N is a tuneable value which defaults to 100 and is set by the "com.ibm.CORBA.threadStateCleanupDelta" property. Test Comments: Circumvention: Don't instantiate new threads within a servlet or manage a thread pool to limit the number of new threads instantiated. Also, clear all instance data before the thread exits. Temporary Fix: ZE FIX ERROR PQ65082 02/09/10 An efix fix is available for 3.5.3. Comments: APAR: PQ56074 Version: 350 Abstract: AUTHORIZATION FAILED FOR / OCCURS WHEN USING COM.IBM.CORBA.LOCALHOST PARAMETER IN ADMIN.CONFIG Error Description: Environment: WebSphere Application Server (WAS) 3.5.4 . Description: With WAS security enabled, when using the com.ibm.CORBA.LocalHost parameter to embed the hostname into IORs instead of the default IP address of the WAS box, trying to start the admin console results in no login prompt and an "Authorization failed for /" message in the tracefile. Any other admin client like XMLConfig also results in the same behavior. Local Fix: None Logically Dependent Apars: Users Affected: WebSphere Application Server users who set-up com.ibm.CORBA.LocalHost parameter in admin.config. Problem Description: With WAS security enabled, when using the com.ibm.CORBA.LocalHost = xxxx, adminserver and/or adminconsole can not be started properly, and you also see authorization failed for /. Recommendation: Apply efix, or not edit com.ibm.CORBA.LocalHost(using default option). Problem Summary: With WAS security enabled, when using the com.ibm.CORBA.LocalHost parameter, adminserver and/or adminconsole can not be started properly, and you also see Authorization failed for / exceptions. Problem Conclusion: With WAS security enabled, when using the com.ibm.CORBA.LocalHost= hostname, the hostname was used in security tag componnent, but IP address was returned by IOR, and the inconsistancy cause TaggedSecurityComponnent not found, thus communication is unsecured. Test Comments: Circumvention: Temporary Fix: PQ56074-354.jar Comments: APAR: PQ57814 Version: 400 Abstract: MACHINE WITH WAS SECURITY STOPS RESPONDING AFTER A PERIOD OF TIME. Error Description: Customer reports running into a problem where a machine configured with WAS security stops working eventually after 5-30mins of working okay. . The machine info is, . OS: SUN DB: Oracle WebServer: IPlanet WAS Version: 4.0.2 . Steps to recreate: 1. Enable OS EJB Security 2. Run normal test cases. . MAchine works for about 5-30 mins. Then "dies". The main symptom seems to be that after logging in to the environment, the user can access the secured site for a period of time, then they suddenly can no longer perform operations on the secured pages and may encounter blank sreens. . This is occuring on WebSphere Commerce Server. Local Fix: None. Logically Dependent Apars: Users Affected: All WebSphere Application Server users of security. Problem Description: Machine with WebSphere Application Server security stops responding after a period of time Recommendation: Problem Summary: After the security is enabled, the application works fine only for a certain period of time. Then the application would fail with the exception JSAS0240E: Login failed. Problem Conclusion: This problem is caused by multiple key file handles opened by WebSphere Application Server. The login failed problem is gone after allowing only one key file to be opened at a time. Test Comments: Circumvention: Temporary Fix: PQ57814_eFix_AEServer.jar Comments: APAR: PQ59150 Version: 350 APAR: PQ59447 Version: 400 Abstract: USERS & GROUPS WHOSE NAME/STRING IS MORE THAN 127 BYTES LONG WILL GET "JAVA.LANG.NEGATIVEARRAYSIZEEXCEPTION" EXCEPTIONS Error Description: Users who belong to more than 127 groups or belong to a group whose name is greater then 127 bytes long will fail authorization. A "NegativeArraySizeException" message will be evident in traces and possibly be visable on a web client. Local Fix: none Logically Dependent Apars: Users Affected: All WebSphere Application Server users using security and adding users to groups with names greater than 127 bytes long or adding users to more than 127 groups. All WebSphere Application Server users using security and adding users to groups with names greater than 127 bytes long or adding users to more than 127 groups. Problem Description: Users encounter an authentication failure for Java clients or an Error 500 for web clients. Recommendation: Problem Summary: Users encounter an authentication failure for Java clients or an Error 500 for web clients. The problem is a NegativeArraySizeException occurs in SecurityAttributeList due to the conversion of byte values into integers before bitwsie operations were performed. Negative byte values were sign extended yeilding a negative value for the final integer after bitwise operations completed. Problem Conclusion: Sign extended bits were masked off before the bitwise operation was performed. Test Comments: Circumvention: Limit users to 127 groups and to groups with less then 128 bytes in their name. Temporary Fix: PQ59150-3.5.5-3.5.6.jar Comments: APAR: PQ65687 Version: 400 Abstract: SECURITY EXCEPTIONS AFTER SERVER REBOOTS Error Description: We are currently having intermittent problems with the IBM WAS Admin service not starting, especially after server reboots. This generates the following entry in sas_server.log: . 1> 2002-07-24 14:23:16.772 , ServerID: 12221999 , SecurityTaggedComponentAssistorImpl.register , Thread Hash: 673378419 Thread P=596272:O=0:CT,5,main : . JSAS0026E: Exception connecting object to the ORB. Check the sas.server.props file to ensure that the SSL keyStore and trustStore properties are set properly. If the problem persists, contact support for assistance. . We did apply the following fix found from the IBM Websphere support site but the problem still recurs. Solution . Put the following parameters in separate lines in the /bin/admin.config and set them to unique unused ports on the machine and restart WAS: . com.ibm.CORBA.SSLPort= com.ibm.CORBA.LSDSSLPort= . Problem occurs on WAS AE 4.03 and 4.04. Local Fix: no workaround available Logically Dependent Apars: Users Affected: All WebSphere Users that have enabled security and restart their adminstration server while their application servers are up and running. Problem Description: If application server is up and running while the adminstration server is restarted, then application server continues to uses a stale IOR without loading the newly created one. Recommendation: Problem Summary: If the WebSphere Application Servers are up and running while the Admin Server is rebooted; then the Application servers continue to use the stale IOR for the Adminstration Server instead of getting the new IOR for the Adminstration Server. With the stale IOR, the Application Servers get OBJECT_NOT_EXIST exceptions. Problem Conclusion: To resolve this problem, the bootstrap repository is read to load the new IORs for the server that was restarted. The logic is to read the bootstrap repository to get the IORs up to three times and ping the server with the new IOR. After three times, if the server does not respond to the new IOR, then OBJECT_NOT_EXIST exception is thrown. Test Comments: Circumvention: Temporary Fix: Comments: APAR: PQ66004 Version: 400 Abstract: MULTIPLE SECURITY LOG ENTRIES FOR ONE INVALID USERID/PASSWORD. Error Description: Upon entering a wrong password for a given userid, the following log statements are written to appserver-out.log. Apart from the fact, that times the same message seems unnecessary, it should the application's decision whether to write or not to write such message regarding invalid logins. Otherwise, a very simple denial of service attack could logins. Otherwise, a very simple denial of service attack could be convinced with bogus login attempts keeping the machine busy logging these messages. 8/26/02 13:17:39:080 CEST 37b2205d SystemOut U 5> 2002-08-26 13:17:39.08 , ServerID: 375334458 , LoginHelperImpl.request_login_controlled : 8/26/02 13:17:39:090 CEST 37b2205d SystemOut U JSAS0240E: Login failed. Verify the userid/password is correct. Check the properties file to ensure the login source is valid. If this error occurs on the server, check the server properties to ensure the principalName has a valid realm and userid. 8/26/02 13:17:39:110 CEST 37b2205d SystemOut U 6> 2002-08-26 13:17:39.11 , ServerID: 375334458 , CredentialsImpl.get_mapped_credentials : 8/26/02 13:17:39:120 CEST 37b2205d SystemOut U JSAS0240E: Login failed. Verify the userid/password is correct. Check the properties file to ensure the login source is valid. If this error occurs on the server, check the server properties to ensure the principalName has a valid realm and userid. 8/26/02 13:17:39:451 CEST 37b2205d SystemOut U 8> 2002-08-26 13:17:39.451 , ServerID: 375334458 , CredentialsImpl.get_mapped_credentials : 8/26/02 13:17:39:481 CEST 37b2205d SystemOut U JSAS0240E: Login failed. Verify the userid/password is correct. Check the properties file to ensure the login source is valid. If this error occurs on the server, check the server properties to ensure the principalName has a valid realm and userid. Action Planned: Sending to entitlement, then to WAS for customer Local Fix: none Logically Dependent Apars: Users Affected: WebSphere Application Server users who have security enabled. Problem Description: Multiple security log entries for one invalid login. Recommendation: Problem Summary: Upon entering a wrong password for a given userid, the following 3 log statements are written to appserver-out.log. It's unnecessary to have 3 login error messages. 8/26/02 13:17:39:080 CEST 37b2205d SystemOut U 5> 2002-08-26 13:17:39.08 , ServerID: 375334458 , LoginHelperImpl.request_login_controlled : 8/26/02 13:17:39:090 CEST 37b2205d SystemOut U JSAS0240E: Login failed. Verify the userid/password is correct. Check the properties file to ensure the login source is valid. If this error occurs on the server, check the server properties to ensure the principalName has a valid realm and userid. 8/26/02 13:17:39:110 CEST 37b2205d SystemOut U 6> 2002-08-26 13:17:39.11 , ServerID: 375334458 , CredentialsImpl.get_mapped_credentials : 8/26/02 13:17:39:120 CEST 37b2205d SystemOut U JSAS0240E: Login failed. Verify the userid/password is correct. Check the properties file to ensure the login source is valid. If this error occurs on the server, check the server properties to ensure the principalName has a valid realm and userid. 8/26/02 13:17:39:451 CEST 37b2205d SystemOut U 8> 2002-08-26 13:17:39.451 , ServerID: 375334458 , CredentialsImpl.get_mapped_credentials : 8/26/02 13:17:39:481 CEST 37b2205d SystemOut U JSAS0240E: Login failed. Verify the userid/password is correct. Check the properties file to ensure the login source is valid. If this error occurs on the server, check the server properties to ensure the principalName has a valid realm and userid. Problem Conclusion: 2 unnecessary messages are removed. Only one message will be logged. Test Comments: Circumvention: Temporary Fix: PQ66004_eFix_test.jar Comments: APAR: PQ66449 Version: 400 Abstract: AUTH PRINCIPAL NULLS OUT AND CAUSES AUTHORIZATION FAILURE. Error Description: Note from Tom D. at the WCC . Trace and this note on wasdoc0. . - error occurs on thread 5f8245 - prior authentication checks on this thread have succeeded - the failure occurs at 13:06:58 - sequence of events: - EJSSecurityCollaborator.preInvoke - received and invoke creds are both non-null - SetUnauthenticatedCredIfNeeded returns false, meaning that the creds are authorized - SecurityCollaborator.performAuthorization is called - since invokedCred is non-null, curReceived will be set to the value of invokedCred - ejbCheckAuthorization is called with curReceived - WSCheckAccessManager.checkAccess is called. One of the args is a new WSPrincipal, which is constructed with the receivedCred. - in checkAccess, creds is initialized to null. If the principal is non-null, then getCredentials will be called on the principal object. - isGrantedAnyRole is called with the creds object. If the creds object is null, or creds.isUnauthenticated is true, then the error occurs. Since there is one error message for two possible conditions, we do not know which one triggered the error. creds can be null if: 1) the principal passed to checkAccess is null. Then creds will not be set. 2) the call to getCredentials on the principal passed in returns null. - not sure where the code is for isUnauthenticated, so I do not know what conditions would cause this call to return true. * but, upon return to checkAccess from isGrantedAnyRole, a message is printed with the value of principal.toString(), which is UNATHENTICATED. So if the principal is UNAUTHENTICATED, then the call to getCredentials will probably produce null. . QUESTIONS: - why does the same thread seem to be able to create ejbs and have the authentication checks succeed, and then later fail on a subsequent create check? - from looking at two different traces, it does not appear as if the failure only occurs on one ejb. So that could rule out a config/deployment issue. - does an UNAUTHENTICATED invokeCreds object somehow get associated with the call? invokedCred is set by a call to current.get_credentials(). . At this point, I'm thinking that the call to current.get_credentials() in performAuthorization is perhaps not returning the proper credentials, and this leads to the client appearing as unauthenticated. But I still do not have enough evidence to back this up. Local Fix: No workaround available. Logically Dependent Apars: Users Affected: WebSphere Application Server users with security enabled and LocalOS as user registry. Problem Description: The authentication principal is lost which causes authorization failure. Recommendation: Problem Summary: Authentication principal becomes unauthenticated when a valid uid/password is used. Actual Exception: ================= securityName: /UNAUTHENTICATED;accessID: UNAUTHENTICATED is not granted any of the required roles: XXXXX 9/24/02 9:23:37:698 EDT 261218 SecurityColla D Authorization failed accessing EJB com.ibm.ws.security.core.AccessException: securityName: /UNAUTHENTICATED;accessID: UNAUTHENTICATED is not granted any of the required roles: Problem Conclusion: LocalOS credentials were not mapped correctly to the OS user registry when a web request qas received. LocalOS credentials are now correctly mapped. Test Comments: Circumvention: Temporary Fix: available Comments: