[
https://issues.apache.org/jira/browse/CASSANDRA-463?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12760629#action_12760629
]
Jonathan Ellis commented on CASSANDRA-463:
------------------------------------------
Thinking about it more: instead of having a FlushPeriodInMinutes per-CF that
could possibly force extra flushes if the CF is busy (filling up and flushing
from normal updates), I suggest adding a global MemtablesFlushAfter setting
(along with SizeInMB and ObjectCountInMillions) which flushes any memtable when
it has been dirty for longer than the given number of minutes. This way
memtables that are constantly being flushed from high activity don't get
"extra" flushes from a misconfigured FlushPeriodInMinutes, and you'll never
have commitlog segments pile up because a CF you thought would always be busy,
didn't fill up for a few days at one point.
> Make FlushPeriodInMinutes default to 60
> ---------------------------------------
>
> Key: CASSANDRA-463
> URL: https://issues.apache.org/jira/browse/CASSANDRA-463
> Project: Cassandra
> Issue Type: Task
> Components: Core
> Reporter: Jonathan Ellis
> Priority: Minor
> Fix For: 0.5
>
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
