23 Aug, 2007

12 commits

  • 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2760 41b1ffd8-f28e-4786-ab96-9950f0a78031
    mkeyes
     
  • As part of https://opends.dev.java.net/issues/show_bug.cgi?id=466 org.opends.admin.ads.ADSContext#registerServer will add an entry for the instance key public-key certificate under 'cn=instance keys,cn=admin data'.
    
    Conversely, the ADSContext#unregisterServer method will deprecate the instance key public-key certificate (i.e., add ds-cfg-key-deprecation-time to the key entry under cn=instance keys,cn=admin data) associated with the server.
    
    Adding or deprecating a instance key public-key certificate entry in ADS has side effects across the ADS domain hosts. In particular, if the instance is an ADS host, each shared secret (encryption) key is wrapped in the instance's public key certificate, so on an addition, the keys are wrapped and added to ADS, while on a certificate deprecation, the keys wrapped with that certificate are deleted.
    
    In the current implementation of org.opends.quicksetup.installer.Installer#updateADS , if the installer detects an ADS entry for the to-be-registered server, it calls ADSContext.unregisterServer then ADSContext.registerServer.  To avoid the perturbation in the ADS-based key distribution, I have replaced the unregister-register sequence with a call to ADSContext.updateServer (which does an LDAP modify replace for the attribute values in the server attribute map). This change was accomplished by calling ADSContext.registerOrUpdateServer.
    
    The changes also include some minor code cleanup, comments, and tidying to eliminate IDEA warnings in ADSContext.
    
    Tests:
    
    I have run the precommit target and done some simple tests to ensure a 1) a remote standalone instance can be promoted to an ADS during an new instance creation; 2) an instance can be created as an ADS and another new instance can be created and added to that ADS; and 3) an instance already in ADS can be destroyed (without unregistering) and recreated and re-added to the ADS (the updateServer scenario). The message for that case is now:
    
    Aug 23, 2007 11:04:05 AM org.opends.quicksetup.installer.Installer updateADS
    WARNING: Server was already registered. Updating server registration.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2759 41b1ffd8-f28e-4786-ab96-9950f0a78031
    david_page
     
  • - MessageDescriptors cache format strings by Locale
    - Check for whether or not the Formatter is required is performed only as needed by MessageDescriptors and not upon Message.toString()
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2758 41b1ffd8-f28e-4786-ab96-9950f0a78031
    kenneth_suter
     
  • …AP connection that can be used to schedule these operations as tasks in addition to the current behavior of operating locally.
    
    - In order to add the LDAP connection arguments I changed the short identifier for the -h/--hash argument of back up to be -A
    
    - BooleanArgument was changed to have an implicit value equal to that of the value of isPresent() and a default value of "false".
    
    - Scheduling an ImportLDIF task now respects the quiet argument and does not print a confirmation message if present.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2757 41b1ffd8-f28e-4786-ab96-9950f0a78031
    kenneth_suter
     
  • 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2756 41b1ffd8-f28e-4786-ab96-9950f0a78031
    mkeyes
     
  • …revious value is longer then the new value. If the previous 
    value was set to a long time (ie 5 hours), the new value wouldn't take effect until the previous 5 hours has expired. With this fix, the new 
    value will take effect immediately by interrupting the rotator thread if its asleep.
    
    Fix for issue 2103
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2755 41b1ffd8-f28e-4786-ab96-9950f0a78031
    boli
     
  • …rd argument which is the number of spaces to indent each wrapped line. There are also delegate wrapText() methods which simulate the old behavior by passing an indentation of 0.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2754 41b1ffd8-f28e-4786-ab96-9950f0a78031
    matthew_swift
     
  • git-svn-id: https://svn.forgerock.org/opendj/trunk@2753 41b1ffd8-f28e-4786-ab96-9950f0a78031
    matthew_swift
     
  • git-svn-id: https://svn.forgerock.org/opendj/trunk@2752 41b1ffd8-f28e-4786-ab96-9950f0a78031
    matthew_swift
     
  • The user friendly names (UFN) are usually derived automatically from the component's "name" and "plural-name" attribute. However, sometimes these names are not that user friendly. For example, the names may be abbreviated forms (e.g. replication -> repl or password policy -> pwp). In these cases it would be nice to be abled to display the full human friendly name in help and documentation.
    
    This change includes an overridden UFN for the global configuration. So we should stop seeing messages in applications like "The Global was updated successfully".
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2751 41b1ffd8-f28e-4786-ab96-9950f0a78031
    matthew_swift
     
  • The total update was failing because the updated server was badly exiting the loop
    of meessage reception when it was receiving a message other than total update messages.
    
    This was caused by a mis-placed return null statement .
    
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2750 41b1ffd8-f28e-4786-ab96-9950f0a78031
    gbellato
     
  • 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2749 41b1ffd8-f28e-4786-ab96-9950f0a78031
    gary_williams
     

22 Aug, 2007

13 commits

  • …ration was incorrectly called "base-level"
    
    Fix for issue 2134
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2748 41b1ffd8-f28e-4786-ab96-9950f0a78031
    boli
     
  • 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2747 41b1ffd8-f28e-4786-ab96-9950f0a78031
    mkeyes
     
  • …tedResult, in the function, searchStringForSubstring.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2746 41b1ffd8-f28e-4786-ab96-9950f0a78031
    mkeyes
     
  • 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2745 41b1ffd8-f28e-4786-ab96-9950f0a78031
    mkeyes
     
  • 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2744 41b1ffd8-f28e-4786-ab96-9950f0a78031
    mkeyes
     
  • …rguments that allow the task to be scheduled to run in the directory server's JVM through the task interface as well as through the existing method or operating within the running JVM.
    
    - This introduces several classes that handle the plumbing of scheduling a task:
    
    LDAPConnectionArgumentParser:  an argument parser that comes pre-populated with the common options used to specify an LDAP connection and includes a method for creating a connection
    
    TaskScheduleInformation:  source of information important for creating the scheduled task entry in the backend
    
    TaskScheduleClient:  interacts on behalf of clients that wish to schedule tasks
    
    TaskTool:  base class that the tools can implement to handle decision about whether to operate locally or remotely
    
    - For export-ldif I removed the short identifier -w for the wrap options since it conflicted with the -w password option.
    
    - For import-ldif I removed the short identifiers -K for skipFile and -q for quiet since they conflicted with the options for specifying the keystore path and startTLS options respectively. 
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2743 41b1ffd8-f28e-4786-ab96-9950f0a78031
    kenneth_suter
     
  • git-svn-id: https://svn.forgerock.org/opendj/trunk@2742 41b1ffd8-f28e-4786-ab96-9950f0a78031
    kenneth_suter
     
  • More refactoring to account for the new parameter, expectedResult, in the searchStringForSubstring function.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2741 41b1ffd8-f28e-4786-ab96-9950f0a78031
    mkeyes
     
  • When using the setup to configure 2 servers in multimaster mode.
    The setup first creates the configuration of the first server and start it
    When the second setup is run to configure the second server, the setup updates
    the configuration of the replication-server on the first server, however the code for handling this dynamic configuration was not yet written and this
    information is not taken into account.
    
    The replication servers are therefore not connected and the replication server
    on the new host therefore does not get the updates.
    
    These changes implement the dynamic configuration of replication servers so that
    the same scenario works fine.
    
    The only properties that is now not dynamically configurable is the
    path of the changelog database.
    
    A new test has been added for these changes.
    I've also tested manually that this does fix the replication server problem
    after configuration using the setup program. 
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2740 41b1ffd8-f28e-4786-ab96-9950f0a78031
    gbellato
     
  • 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2739 41b1ffd8-f28e-4786-ab96-9950f0a78031
    gary_williams
     
  • 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2738 41b1ffd8-f28e-4786-ab96-9950f0a78031
    gary_williams
     
  • The replication monitoring information already provides the
    attribute unresolved-naming-conflicts that counts the number of conflicts
    that was not automatically resolved since last startup.
    
    The conflicting entries are also marked with the ds-sync-confict attribute
    so that administrators can look for these entries 
    using filter ds-sync-confict=*
    
    This code add the generation of an administrative alert when a conflict is
    detected so that administrators can be made aware of the problem.
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2737 41b1ffd8-f28e-4786-ab96-9950f0a78031
    gbellato
     
  • Before this fix, If one remove the root entry of a replication domain,
    then stop the server The PersistentServerState is lost.
    Therefore when later restarting the server all the changes from
    the replication server are replayed on the database.
    
    To avoid this the fix is simply to store the PersistentServerState to the configuration entry when the root entry does not exist.
    
    I have not developed a unit test for this scenario because I believe that it is not possible to restart the server during the unit test.
    I have done manual tests to check that the PersistentServerState
    is correctly saved and reread. 
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@2736 41b1ffd8-f28e-4786-ab96-9950f0a78031
    gbellato
     

21 Aug, 2007

15 commits