accumulo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eric Newton <eric.new...@gmail.com>
Subject Re: How to delete all rows in Accumulo table?
Date Tue, 15 Apr 2014 13:37:13 GMT
You can get the splits (into a file) and restore them (from a file).

I can give you one use-case where you wish to lose your splits.
Consider indexing information based on the current date.  And you want
to remove old data.  That is, you will never put old data back into
those old splits.  It makes sense to remove the rows, and the splits
associated with them.

deleterows is a sledgehammer.  It is not a fine tool for removing a
few stray elements.  Its a way of efficiently removing a month of data
from your 13 months of records that is (partially) indexed on month.

-Eric

On Mon, Apr 14, 2014 at 8:50 PM, Arshak Navruzyan <arshakn@gmail.com> wrote:
> BTW, I noticed that with
>
> deleterows -t foo -f
>
> you lose your split points.  Not sure why this is desirable behavior in the
> code.
>
>
> On Mon, Apr 14, 2014 at 12:53 PM, Tiffany Reid <treid@eoir.com> wrote:
>>
>> Thanks so much for all responses.
>>
>> Tiffany
>>
>> On Apr 14, 2014, at 3:24 PM, "Keith Turner" <keith@deenlo.com> wrote:
>>
>> deletemany pulls data back to the client and write deletes back.   The
>> deleterows command is more efficient, it preforms operations on the server
>> side.  Entire tablets that fall within the range are just dropped.
>>
>>
>>
>>
>> On Mon, Apr 14, 2014 at 3:20 PM, Russ Weeks <rweeks@newbrightidea.com>
>> wrote:
>>>
>>> deletemany -t <table> -f
>>>
>>> If you have a large table, that command will produce a lot of output. I
>>> don't know if there's a way to make it less verbose? Maybe best to pipe it
>>> to /dev/null.
>>>
>>> -Russ
>>>
>>>
>>> On Mon, Apr 14, 2014 at 12:13 PM, Tiffany Reid <treid@eoir.com> wrote:
>>>>
>>>> Hi,
>>>>
>>>>
>>>>
>>>> How do I delete all rows in a table via Accumulo Shell?
>>>>
>>>>
>>>>
>>>> Thanks,
>>>>
>>>> Tiffany
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>

Mime
View raw message