[
https://issues.apache.org/jira/browse/CASSANDRA-11369?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15213416#comment-15213416
]
Robert Stupp commented on CASSANDRA-11369:
------------------------------------------
Unlike the initial approach, CASSANDRA-11206 now distinguishes between RIEs
that keep {{IndexInfo}} objects on-heap and those that are too big and are
never fully materialized on heap. In other words: RIEs with index samples and a
serialized size up to {{column_index_cache_size_in_kb}} behave as before 11206.
> Re-use IndexedEntry / IndexInfo after 11206
> -------------------------------------------
>
> Key: CASSANDRA-11369
> URL: https://issues.apache.org/jira/browse/CASSANDRA-11369
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Robert Stupp
> Assignee: Stefania
> Priority: Minor
> Fix For: 3.x
>
>
> CASSANDRA-8180 introduced
> {{UnfilteredRowIteratorWithLowerBound#getPartitionIndexLowerBound}} which
> relies on {{IndexInfo}} being present in the key-cache. But CASSANDRA-11206
> will remove the presence of {{IndexInfo}} objects in-memory.
> This ticket is about to re-add that functionality if feasible.
> Currently tagged with fix-version 3.x, if it is possible to to that without
> changing summary or index disk structures - but maybe it's necessary to
> change the on-disk structure, which means we can do this in 4.x.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
