[
https://issues.apache.org/jira/browse/OAK-12005?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18037474#comment-18037474
]
Julian Sedding commented on OAK-12005:
--------------------------------------
The number of UUIDs after the fix to deduplicate them came down from ~20mio to
~1mio. The memory consumption dropped to ~36MB for {{UUID}} and next to nothing
for {{HashMap$Node}}.
> segment preloading graph-cache uses too much heap
> -------------------------------------------------
>
> Key: OAK-12005
> URL: https://issues.apache.org/jira/browse/OAK-12005
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: segment-tar
> Affects Versions: 1.88.0
> Reporter: Julian Sedding
> Assignee: Julian Sedding
> Priority: Major
> Fix For: 1.90.0
>
>
> The {{SegmentPreloader}} added with OAK-11934 caches the segment-graph of
> each archive. This leads to a large number of UUID instances, consuming lots
> of heap memory.
> On a segmentstore with ~200 archives I observed 648MB of heap space occupied
> by 20'261'269 UUID instances, together with another 645MB occupied by
> 20'161'308 {{HashMap$Node}} instances.
> The UUID instances should be deduplicated to reduce the memory footprint.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
