Yes, that's right. And that's why you can compact ranges in 1.5.
In 1.6.0, the compaction strategy will be pluggable so you can do this sort
of compaction automatically.
On Wed, Nov 6, 2013 at 5:07 PM, Dickson, Matt MR <
[EMAIL PROTECTED]> wrote:
> We have data ingested in a date partitioned table, ie splits are created
> for each day in the form yyyymmdd-nnn where n is a random number appended
> to ensure the data is spread across the cluster. On the table we have
> added an ageoff filter with scan, minc and majc scopes and we load the
> table using bulk ingest. Because older splits have no new data going into
> them it appears no major compactions occur so the ageoff filter is not
> removing old data. I suspect the way to force a majc is to merge old
> tablets is this correct?
> Thanks in advance,
> Matt Dickson