Jimmy Xiang created HBASE-6745:
----------------------------------
Summary: Always flush region based on memstore size could hold
hlog files from archiving
Key: HBASE-6745
URL: https://issues.apache.org/jira/browse/HBASE-6745
Project: HBase
Issue Type: Bug
Components: replication, wal
Reporter: Jimmy Xiang
Currently, memstore flusher always chooses the biggest memstore region to flush.
Suppose I have two tables: one is very actively updated, while the other is
periodically updated. The active one has biggest memstore all the time and is
flushed all the time. But the in-active one never gets a chance to flush.
Since it is not flushed, the hlog file can't be archived, although there are
lots of hlog files.
If the active table happens to have big updates all the time, the hlog files
could cause huge disk space pressure.
Other than the memstore size, periodically flushing regions based on hlog roll
time is helpful in hlog archiving/replication.
--
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
