hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew Purtell (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HBASE-1733) Pluggable replication framework
Date Sat, 07 Nov 2009 04:05:41 GMT

     [ https://issues.apache.org/jira/browse/HBASE-1733?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Andrew Purtell updated HBASE-1733:
----------------------------------

    Description: 
Put minimal support for replication into core and plan for the rest in contrib/. Support in
core can look like:

1. Scope tags in WAL and HCD (HBASE-1728, HBASE-1834). Generic integer, sufficient for any
number of schemes.

and handled in this issue:

2. Method to register remote watchers on WAL rolls, via watcher in ZK. So this is basically
distributed upcall to the listeners on roll. 

3. Mechanism to prevent HLog from cleaning rolled logs until all listeners have finished with
it, also implemented with ZK. Could only delete when no more ephemeral subnodes on the log
node, where region server would hold an entry also and delete its entry when edit content
has made it out to the storefiles. 
 

  was:
Put minimal support for replication into core and plan for the rest in contrib/. Support in
core can look like:

# Scope tags in WAL and HCD (HBASE-1728, HBASE-1834). Generic integer, sufficient for any
number of schemes.

and handled in this issue:

# Method to register remote watchers on WAL rolls, via watcher in ZK. So this is basically
distributed upcall to the listeners on roll. 

# Mechanism to prevent HLog from cleaning rolled logs until all listeners have finished with
it, also implemented with ZK. Could only delete when no more ephemeral subnodes on the log
node, where region server would hold an entry also and delete its entry when edit content
has made it out to the storefiles. 
 


> Pluggable replication framework
> -------------------------------
>
>                 Key: HBASE-1733
>                 URL: https://issues.apache.org/jira/browse/HBASE-1733
>             Project: Hadoop HBase
>          Issue Type: Sub-task
>            Reporter: Andrew Purtell
>             Fix For: 0.21.0
>
>
> Put minimal support for replication into core and plan for the rest in contrib/. Support
in core can look like:
> 1. Scope tags in WAL and HCD (HBASE-1728, HBASE-1834). Generic integer, sufficient for
any number of schemes.
> and handled in this issue:
> 2. Method to register remote watchers on WAL rolls, via watcher in ZK. So this is basically
distributed upcall to the listeners on roll. 
> 3. Mechanism to prevent HLog from cleaning rolled logs until all listeners have finished
with it, also implemented with ZK. Could only delete when no more ephemeral subnodes on the
log node, where region server would hold an entry also and delete its entry when edit content
has made it out to the storefiles. 
>  

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message