[
https://issues.apache.org/jira/browse/HBASE-4071?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Lars Hofhansl updated HBASE-4071:
---------------------------------
Attachment: MinVersions.diff
"A patch is worth a thousand words."
Here's an *idea* for patch.
Still getting familiar with the HBase code (so please cut me some slack, I may
have missed an entire subsystem.)
I did some light testing both with Wildcard and Explicit Column Trackers, and
things seem to work for me so far, including the shell.
Not too happy about the patch, though:
1. Expired rows can no longer be expunged when a Store's internal cache is
flushed (and minversions > 0). Although I don't see how that can be avoided as
not all versions are in the cache.
2. Makes the code harder to follow.
3. I just stubbed out the testing stuff. For now.
4. It's hardcoded (Todd won't like it :) )
5. Passes a MatchCode to ColumnTracker.checkColumn
6. The relation between TTL and Version just became more complicated.
7. Can I assume the Trackers deliver KV's in reverse time order?
Please let me know if this is off track.
Thanks...
> Data GC: Remove all versions > TTL EXCEPT the last written version
> ------------------------------------------------------------------
>
> Key: HBASE-4071
> URL: https://issues.apache.org/jira/browse/HBASE-4071
> Project: HBase
> Issue Type: New Feature
> Reporter: stack
> Attachments: MinVersions.diff
>
>
> We were chatting today about our backup cluster. What we want is to be able
> to restore the dataset from any point of time but only within a limited
> timeframe -- say one week. Thereafter, if the versions are older than one
> week, rather than as we do with TTL where we let go of all versions older
> than TTL, instead, let go of all versions EXCEPT the last one written. So,
> its like versions==1 when TTL > one week. We want to allow that if an error
> is caught within a week of its happening -- user mistakenly removes a
> critical table -- then we'll be able to restore up the the moment just before
> catastrophe hit otherwise, we keep one version only.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
