[jira] [Commented] (HBASE-14272) Enforce major compaction on stores with TTL or time.to.purge.deletes enabled

["Lars Hofhansl (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-14272?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14735863#comment-14735863 ] 

Lars Hofhansl commented on HBASE-14272:
---------------------------------------

OK... Assigned to me.

Let's simply start with always doing the time based compaction when TTL is set. Period. Simple.

Then we do the following (sub jira or here):
# track a flag during compactions that is set for any HFile that has any delete markers in it.
# track the maximum number of version we have seen for any Column. Note: need to make sure that we can do that cheaply (need a compare of row-key, family, CQ).

With this in place we should be able to decide whether a compaction by the following:
{code}
...
} else if (TTL set) {
  if (MIN_VERSIONS > 0) {
    if (store.maxNumVersions() > MIN_VERSIONS) {
       compact();
    }
  } else {
    compact()
  }
} else if (time.to.purge.deletes > 0) {
   if (store.hasDeletes() && !KEEP_DELETED_CELLS) {
     compact()
   }
}
{code}

I'm sure this logic can be expressed more nicely, but you get the point.


> Enforce major compaction on stores with TTL or time.to.purge.deletes enabled
> ----------------------------------------------------------------------------
>
>                 Key: HBASE-14272
>                 URL: https://issues.apache.org/jira/browse/HBASE-14272
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Vladimir Rodionov
>            Assignee: Lars Hofhansl
>             Fix For: 2.0.0
>
>         Attachments: HBASE-14272-v2.patch, HBASE-14272.patch
>
>
> Currently, if store has one (major compacted) file, the only case when major compaction will be triggered for this file again - when locality is below threshold, defined by *hbase.hstore.min.locality.to.skip.major.compact* or TTL expired some cells. If file has locality greater than this threshold it will never be major compacted until Store's TTL kicks in. For CF with KEEP_DELETED_CELLS on, compaction must be enabled always (even for single file), regardless of locality, when deleted cells are expired (*hbase.hstore.time.to.purge.deletes*)



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