[jira] [Updated] (CASSANDRA-4778) leveled compaction does less work in L0 than intended

["Jonathan Ellis (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/CASSANDRA-4778?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jonathan Ellis updated CASSANDRA-4778:
--------------------------------------

    Attachment: 4778.txt

patch to move combine-with-L1 logic outside the candidates loop
                
> leveled compaction does less work in L0 than intended
> -----------------------------------------------------
>
>                 Key: CASSANDRA-4778
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-4778
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Jonathan Ellis
>            Priority: Minor
>              Labels: compaction
>         Attachments: 4778.txt
>
>
> We have this code in the candidate loop:
> {code}
> .               if (SSTable.getTotalBytes(candidates) > maxSSTableSizeInBytes)
>                 {
>                     // add sstables from L1 that overlap candidates
>                     candidates.addAll(overlapping(candidates, generations[1]));
>                     break;
>                 }
> {code}
> thus, as soon as we have enough to compact to make one L1 sstable's worth of data, we stop collecting candidates.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira