[
https://issues.apache.org/jira/browse/SOLR-6331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14095575#comment-14095575
]
Brett Lucey commented on SOLR-6331:
-----------------------------------
I think that should be feasible. Previously, we re-validated each level as we
refined. Now we mark a level as complete. Once we know it's complete, I think
we should be able to trim during refinement as you suggest and release some of
the memory. I definitely think it's worth getting SOLR-2894 committed before
looking at this optimization though, and I assume you agree since you separated
this out into a separate Jira issue?
> possible memory optimization for distributed pivot faceting
> -----------------------------------------------------------
>
> Key: SOLR-6331
> URL: https://issues.apache.org/jira/browse/SOLR-6331
> Project: Solr
> Issue Type: Improvement
> Reporter: Hoss Man
>
> As noted in a comment in {{PivotFacetField.trim()}}...
> {code}
> // we can probably optimize the memory usage by trimming each level of the
> pivot once
> // we know we've fully refined the values at that level
> // (ie: fold this logic into refineNextLevelOfFacets)
> {code}
--
This message was sent by Atlassian JIRA
(v6.2#6252)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
