[jira] [Commented] (CASSANDRA-4474) Respect five-minute flush moratorium after initial CL replay

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

    [ https://issues.apache.org/jira/browse/CASSANDRA-4474?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13447997#comment-13447997 ] 

Jonathan Ellis commented on CASSANDRA-4474:
-------------------------------------------

I think the (existing) "is compaction manager idle" check is bogus: if we have enough CFs that CM starts working on the first, before it finishes looping over everyone, then some CFs will never have compaction kicked off.  Suggest just blindly submitting the background compaction, which will no-op if it's redundant (which is what flush does as well).

Otherwise LGTM.
                
> Respect five-minute flush moratorium after initial CL replay
> ------------------------------------------------------------
>
>                 Key: CASSANDRA-4474
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-4474
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>            Reporter: Jonathan Ellis
>            Assignee: Yuki Morishita
>            Priority: Minor
>              Labels: commitlog, compaction
>             Fix For: 1.1.5
>
>         Attachments: 4474-1.1.txt
>
>
> As noted in CASSANDRA-1967, the post-replay flush can kick off compactions before the five minute grace period introduced in CASSANDRA-3181 to avoid i/o contention while server is warming up.

--
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