[
https://issues.apache.org/jira/browse/OAK-4451?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15331925#comment-15331925
]
Michael Dürig commented on OAK-4451:
------------------------------------
I pushed an initial implementation of the to my private GitHub repository:
https://github.com/mduerig/jackrabbit-oak/commit/0037b886ff8ca5e5945ad37d56d3b8ab084c36f9
[~frm], [~alex.parvulescu], kindly review.
There are some loose ends to tie still: the calculations for entry weights and
sizes in the [TemplateCache |
https://github.com/mduerig/jackrabbit-oak/commit/0037b886ff8ca5e5945ad37d56d3b8ab084c36f9#diff-ae41fca59d0da3f5561baf8df3a72631R22]
need still to be done. Maybe [~tmueller] could help out here as he was the
initial implementer of this cache.
The [TemplateCacheTest |
https://github.com/mduerig/jackrabbit-oak/commit/0037b886ff8ca5e5945ad37d56d3b8ab084c36f9#diff-545902106810ed9f3d67b331485c8073R22]
needs some more tests (currently 0 ;-) )
> Implement a proper template cache
> ---------------------------------
>
> Key: OAK-4451
> URL: https://issues.apache.org/jira/browse/OAK-4451
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: segment-tar
> Reporter: Michael Dürig
> Labels: cache, monitoring, production
> Fix For: 1.6
>
>
> The template cache is currently just a map per segment. This is problematic
> in various ways:
> * A segment needs to be in memory and probably loaded first only to read
> something from the cache.
> * No monitoring, instrumentation of the cache
> * No control over memory consumption
> We should there for come up with a proper template cache implementation in
> the same way we have done for strings ({{StringCache}}) in OAK-3007.
> Analogously that cache should be owned by the {{CachingSegmentReader}}.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
