That's great! The 3rd point seems exactly what happened on my column family.
As supposed it was a matter of ignorance :D

Many many thanks
Andrea

On 02/28/2013 01:39 PM, Víctor Hugo Oliveira Molinar wrote:
Let me know if it help you, http://www.datastax.com/docs/1.0/dml/about_writes#about-deletes


On Thu, Feb 28, 2013 at 5:35 AM, Andrea Gazzarini <andrea.gazzarini@gmail.com> wrote:
Hi,
I'm using Cassandra 1.1.8 and today I saw in my keyspace a column family with the following content

> SELECT * FROM challenge;
 KEY
----------------------------
 49feb2000100000a556522ed68
 49feb2000100000a556522ed74
 49feb2000100000a556522ed7a
 49feb2000100000a556522ed72
 49feb2000100000a556522ed76
 49feb2000100000a556522ed6a
 49feb2000100000a556522ed70
 49feb2000100000a556522ed78
 49feb2000100000a556522ed6e
 49feb2000100000a556522ed6c

So, only rowkeys.
Yesterday those rows were there and I ran some deletions (exactly on those rows). I'm using Hector

Mutator<byte []> mutator = HFactory.createMutator(keyspace, BYTES_ARRAY_SERIALIZER)
    .addDeletion(challengeRowKey(...), CHALLENGE_COLUMN_FAMILY_NAME)
    .execute();

This is a small development and test environment on a single machine / single node so I don't believe the hardware details are relevant.
Probably I'm doing something stupid or I didn't get the point about how things are working, but as far I understood the rows above are no valid... column name and column value coordinates are missing so there are no valid cells (rowkey / column name / column value)...is that right?

I read about ghost reads but I think this is a scenario in a distribuited environment...is that valid after one day and on a single Cassandra node??

Regards and thanks in advance,
Andrea