accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Christopher Tubbs (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-2579) Modify "live ingest" code path to create replication entries
Date Thu, 30 Oct 2014 20:34:38 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-2579?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14190787#comment-14190787
] 

Christopher Tubbs commented on ACCUMULO-2579:
---------------------------------------------

This issue is marked as "fixed", but no work is logged on the issue in JIRA (though I do see
commits against it), and no fixVersion specified.

> Modify "live ingest" code path to create replication entries
> ------------------------------------------------------------
>
>                 Key: ACCUMULO-2579
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-2579
>             Project: Accumulo
>          Issue Type: Sub-task
>          Components: replication
>            Reporter: Josh Elser
>            Assignee: Josh Elser
>
> When new writes come in to a server, we want to track the updates to the WAL, and ensure
that we create entries to track that we need to replicate that data.
> Supporting replication over portions of a WAL is desirable to reduce replication latency,
but this may decrease the total ingest throughput.
> Despite not wanting to persist a replication record for every update, but we might need
to so we can ensure that we don't miss data that needs replication. Need to observe how things
like the tserver processes BatchWriter updates -- I believe the problem is analogous.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message