[
https://issues.apache.org/jira/browse/CASSANDRA-856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12842359#action_12842359
]
Jonathan Ellis commented on CASSANDRA-856:
------------------------------------------
this is because the flush code calls switch() on the normal memtable, meaning
flush and replace w/ a new one, but only flush() on the binary one, not
replacing it w/ a new one
> Binary Memtable interface writes the same data after each flush.
> ----------------------------------------------------------------
>
> Key: CASSANDRA-856
> URL: https://issues.apache.org/jira/browse/CASSANDRA-856
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Environment: x86 64-bit CentOS 5.4
> Reporter: Wing Lian
> Priority: Minor
> Fix For: 0.6
>
>
> Subsequent flushes to an a Binary Memtable that has not been modified since
> the last flush has ghost data written out to the data directory. I've
> verified that the md5sum of the all the subsequent files are identical.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
