Summary

Directory synchronization from an external LDAP directory is not successfully synchronizing all users. 

Diagnosis

The following appears in the  atlassian-confluence.log

2018-03-23 18:13:04,980 ERROR [Caesium-1-1] [engine.jdbc.spi.SqlExceptionHelper] logExceptions ERROR: duplicate key value violates unique constraint "cwd_user_name_dir_id"
  Detail: Key (lower_user_name, directory_id)=(<username>, 3276801) already exists.

2018-03-23 18:13:05,055 ERROR [Caesium-1-1] [crowd.embedded.hibernate2.HibernateUserDao] addAll The following users could not be processed:
2018-03-23 18:13:05,056 ERROR [Caesium-1-1] [crowd.embedded.hibernate2.HibernateUserDao] addAll aliceb
2018-03-23 18:13:05,056 ERROR [Caesium-1-1] [crowd.embedded.hibernate2.HibernateUserDao] addAll christinej
2018-03-23 18:13:05,056 ERROR [Caesium-1-1] [crowd.embedded.hibernate2.HibernateUserDao] addAll Please try to resolve any errors with these users, and try again.

Cause

This can be caused by a user existing in multiple OUs in LDAP and more than one of the OUs are being pulled into the user directory in Confluence. Once the duplicated user is hit, directory synchronization for the remaining users is not completed.

Workaround

To resolve the issue eliminate the duplicate from being pulled into Confluence by:

1. Redefine the scope of the LDAP directory in Confluence to exclude one of the OUs so that it does not pull the user twice
2. Or remove the user from one of the duplicated OUs
3. Or create multiple user directories making sure that the set of users being retrieved in each user directory is unique.

We have a feature request open for allowing synchronization to continue after skipping the duplicate:  CONFSERVER-55246 - Getting issue details... STATUS