incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonathan Ellis <jbel...@gmail.com>
Subject Re: Deleted rows showing up when doing a get_range_slice query
Date Wed, 24 Feb 2010 21:25:31 GMT
On Wed, Feb 24, 2010 at 3:21 PM, Erik Holstad <erikholstad@gmail.com> wrote:
> When deleting rows from a table and then using a get_range_slice query, the
> keys or the
> deleted rows show up, with no name/value pairs. What is the reasoning behind
> this?

Cassandra doesn't know that there are no other columns associated with
that key, without doing a more expensive query.  This is what
get_key_range used to do, but it was too slow.  So just treat it as a
row that didn't happen to have the particular columns you requested,
which is what it is, until it gets GC'd.  (See:
http://wiki.apache.org/cassandra/DistributedDeletes)

> I have also seen a weird issue when using a md5 generated byte[] as a column
> name,
> doesn't seem like it actually work. I can't get the value that was inserted
> that way.
> But if I for example Base64.encode().getBytes() it seems to be ok, any
> ideas?

Something is wrong on your end, but you haven't given us enough
details to tell what it is. :)

-Jonathan

Mime
View raw message