Marcel Reutegger created OAK-5010:
-------------------------------------
Summary: Document split with binary properties too eager
Key: OAK-5010
URL: https://issues.apache.org/jira/browse/OAK-5010
Project: Jackrabbit Oak
Issue Type: Bug
Components: core, documentmk
Affects Versions: 1.5.1
Reporter: Marcel Reutegger
Assignee: Marcel Reutegger
Priority: Minor
Fix For: 1.6
OAK-2392 changed the split logic for documents and creates a split document as
soon as a binary property is overwritten. This happens quite frequently in a
default Oak setup with oak-lucene. Lucene files are stored in the repository
and many of them are short lived because they are merged into bigger segments
later. This means all documents for those files are split as soon as the file
is deleted. The revision garbage collection does not get any benefit from this,
because it will remove the files anyway. It is actually more expensive, because
the GC also needs to collect the split documents.
The split logic should be updated to only create a split document with binary
properties when the node still exists at the current head revision.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
