accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Josh Elser (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-2799) Create ReplicationOperations
Date Wed, 14 May 2014 16:15:16 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-2799?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13997711#comment-13997711
] 

Josh Elser commented on ACCUMULO-2799:
--------------------------------------

I'm on to my second implementation of a "drain" operation which a client can invoke to know
"when the data has been replicated" for a table. There are a couple of caveats here that make
this difficult.

* You might get a new WAL for a table even if no data has been ingested into it (WALs are
per-tserver, not per-table)
* How do you implement it with constant ingestion? Prevent the user from making a call that
will never complete. (Is that intended?)

What I've done for the present is take a snapshot of the current WALs in use by a table (using
the metadata table and the 'log' colfam) and wait for all of those WALs to be fully replicated.
This at least addresses both of the above points, but I'm still not positive if this is the
"best" semantics for it. Thoughts, anyone else?

> Create ReplicationOperations
> ----------------------------
>
>                 Key: ACCUMULO-2799
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-2799
>             Project: Accumulo
>          Issue Type: Sub-task
>            Reporter: Josh Elser
>            Assignee: Josh Elser
>             Fix For: 1.7.0
>
>
> It would be nice to make a {{Connector#replicationOperations}} that has some helpful
methods regarding replication.
> * Wait for a table to be fully replicated
> * ... others?



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message