You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Jonathan Hsieh (JIRA)" <ji...@apache.org> on 2013/01/19 16:36:13 UTC
[jira] [Resolved] (HBASE-7291) Simple Log Roll Snapshot
[ https://issues.apache.org/jira/browse/HBASE-7291?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jonathan Hsieh resolved HBASE-7291.
-----------------------------------
Resolution: Won't Fix
Assignee: Jonathan Hsieh
I'm going to close this approach out for the time being -- the flush snapshot is simpler, seems to work fairly robustly, and relatively quick.
> Simple Log Roll Snapshot
> ------------------------
>
> Key: HBASE-7291
> URL: https://issues.apache.org/jira/browse/HBASE-7291
> Project: HBase
> Issue Type: Sub-task
> Reporter: Jonathan Hsieh
> Assignee: Jonathan Hsieh
> Attachments: hbase-7291.v1.patch, pre-hbase-7291.v1.patch, snapshot-online-test-demo.jruby
>
>
> This snapshot will likely be the cheapest snapshot taking method -- it rolls or drops a marker in the hlog when a snapshot request is made. This approach can be seen as an improvement upon copytable/export with similar consistency guarantees. Specifically, there will be "ragged edges", and happens-before relation on clients may not be consistent from a client's point of view but it will be useable for many backup for disaster recovery use cases.
> For example:
> If client 1 writes row a to region A and then row b to region B, a client happens-before enforcing snapshot would only be allowed to have neither a nor b, only a, or both a and b. Having only b should not be possible. This outcome would be possible with the log-roll snapshot.
--
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