[ https://issues.apache.org/jira/browse/HBASE-7680?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sergey Shelukhin updated HBASE-7680: ------------------------------------ Attachment: HBASE-7680-v4-with-7679.patch HBASE-7680-v4.patch The heuristic used to determine whether it's worth it to add L0 to stripe to drop deletes is really stupid, what ends up happening is L0 gets compacted with priority when there's a lot of it, and when there's like one L0 file and lots of data overall, this L0 file always gets split into tiny files when some stripe is compacting. Change to add L0 to drop deletes only when L0 would have compacted anyway otherwise. > implement compaction policy for stripe compactions > -------------------------------------------------- > > Key: HBASE-7680 > URL: https://issues.apache.org/jira/browse/HBASE-7680 > Project: HBase > Issue Type: Sub-task > Reporter: Sergey Shelukhin > Assignee: Sergey Shelukhin > Attachments: HBASE-7680-v0.patch, > HBASE-7680-v0-with-7679-and-7935.patch, HBASE-7680-v1.patch, > HBASE-7680-v1-with-7679.patch, HBASE-7680-v2.patch, > HBASE-7680-v2-with-7679-and-8034.patch, HBASE-7680-v3.patch, > HBASE-7680-v3-with-7679.patch, HBASE-7680-v4.patch, > HBASE-7680-v4-with-7679.patch > > -- 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