[OAK-4564] Use Lucene index to look for long-named nodes in JCR2 upgrade

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.4.12, 1.5.6, 1.6.0
• Component/s: upgrade
• Assignee: Unassigned
• Reporter: Tomek Rękawek
• Created: 15/Jul/16 10:05
• Updated: 16/Dec/16 11:15
• Resolved: 15/Jul/16 10:34

Description

oak-upgrade should perform a check for presence of node with long names. One approach as suggested by mreutegg is to perform this check via Lucene index

we could use the lucene index of the source repository. the node name
is indexed and a scan of those terms with a length check would be
quite efficient.

The expected behavior can be

1. oak-upgrade starts and constructs the JR2 instances
2. It performs the node name check and collects such node names.
3. If the above list is not empty it terminates the migration providing details about the problamatic paths
4. User then has an option to
   1. Restart the migration with --skip-name-check
   2. Fix the content and then migrate

Issue Links

• breaks: OAK-4842 Upgrade breaks if there's no SearchManager configured in repository.xml
• Is contained by: OAK-5290 Backport the performance improvements for oak-upgrade from trunk