Could not replicate the entire batch using the BatchProcessor

[William Wells]

I am cloning a production server for a staging, test with 4.1.2 to start a major upgrade of all of our Atlassian tools. I followed the instructions in https://confluence.atlassian.com/display/CROWD020/Migrating+Crowd+between+Servers and everything seemed to be working fine until I got to the import portion. It starts off fine, but as it is importing I get the error in my subject line. I'm using MySQL for both of the backend databases for test and production. The only difference between the two systems is that the original seems to have been installed as a WAR and the test is a standalone.

[William Wells · Accepted solution]

The problem I found was there was a duplicate user account created with the same user login. The account was for the same person, and it should have been identifiable by Crowd as a separate entity due to the assignment of a uuid to each account, but it didn't. I went into the backup xml file and removed one of the instances completely and was able to move forward in the import process.

[David Feig]

I had the same experience. I manually deleted duplicate users from the xml backup file. These were breaking database unique constraints. I just don't understand how these got into the backup.

I used the following bash command to find dups:

grep '<name>' atlassian-crowd-2.3.4-backup-2012-08-22-200326.xml.edited|sort|uniq -d|less

This finds more than just users but it was pretty easy to tell the wheat from the chaff.

[JustinK]

Will,

Could you provide the complete stacktrace here so we could get a little more context 'in code' of what the problem is during the upgrade process.

Cheers,

Justin

[David Feig]

I had the same experience. I manually deleted repeat users. These were breaking database unique constraints. I just don't understand how these got into the backup.