[
https://issues.apache.org/jira/browse/OAK-2595?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Michael Dürig resolved OAK-2595.
--------------------------------
Resolution: Fixed
Fixed in trunk at http://svn.apache.org/r1665275
Merged into 1.0 at http://svn.apache.org/r1665298.
The fix makes use of a memory optimised {{RecordIdMap}}, which groups record
ids by segment id and stores record offsets as a packed array of shorts along
with them.
With this fix I've seen significant lower memory pressure and churn when
running the estimation process on large repositories.
> High memory consumption of CompactionGainEstimate
> -------------------------------------------------
>
> Key: OAK-2595
> URL: https://issues.apache.org/jira/browse/OAK-2595
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: segmentmk
> Reporter: Michael Dürig
> Assignee: Michael Dürig
> Labels: compaction, gc
> Fix For: 1.1.8, 1.0.13
>
>
> {{CompactionGainEstimate}} keeps a set for the visited record ids. Each entry
> in that set is represented by an instance of {{ThinRecordId}}. For big
> repositories the instance overhead lead to {{OOME}} while running the
> compaction estimator.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
