Importing Crowd XML Backup Fails due to Different Locale
Symptoms
The following error appears in log when importing a Crowd XML backup:
ERROR [atlassian.crowd.migration.DirectoryMapper] Failed to parse date: Wed May 14 18:03:35 +0200 2008
java.text.ParseException: Unparseable date: "Wed May 14 18:03:35 +0200 2008"
at java.text.DateFormat.parse(DateFormat.java:335)
at com.atlassian.crowd.migration.GenericMapper.getDateFromFormattedString(GenericMapper.java:99)
org.springframework.dao.DataIntegrityViolationException: not-null property references a null or transient value: com.atlassian.crowd.model.directory.DirectoryImpl.createdDate; nested exception is org.hibernate.PropertyValueException: not-null property references a null or transient value: com.atlassian.crowd.model.directory.DirectoryImpl.createdDate
at org.springframework.orm.hibernate3.SessionFactoryUtils.convertHibernateAccessException(SessionFactoryUtils.java:665)
.
.
.
Caused by: org.hibernate.PropertyValueException: not-null property references a null or transient value: com.atlassian.crowd.model.directory.DirectoryImpl.createdDate
at org.hibernate.engine.Nullability.checkNullability(Nullability.java:101)
Cause
The JVM locale used to create the date-time data for the XML backup file is different from the one defined in the JVM used to receive the data (where the new Crowd installation is running).
Resolution
Follow these steps to check the previous machine's locale and then set it correctly for the new server's JVM:
- Identify what format/language is being used for the data-time data in the backup file. In the example above, the locale is obviously English/American. Sometimes, running command
localeon the server where the backup was performed can help. Set the JAVA_OPTS variable in Crowd server, with the new language and country from step (1). For example:
export JAVA_OPTS="-Duser.language=en -Duser.country=US"
Last modified on Aug 15, 2014
Powered by Confluence and Scroll Viewport.