legacy transition with get_range_by_token

[fort...@gmail.com]

Hi,

I am trying to move a 18T legacy Compact storage table to CQL3 style keyspace on a 16-node 2.0.11 cluster.  The original table in question has both static and dynamic columns which breaks the use of Spark for this task - Datastax driver only sees the static columns.  

Since I have all the tokens ranges via node tool for  my table, I thought I could use your get_range_by_token in a loop to grab 1000 at a time and reload them into a separate cluster.  My problem is that I can't seem to make the 2nd iteration start from the end of the first since the last row of the original get_range_by_token only has getKey().  I tried iterating like it the following:

```

while($x = $cf->get_range_by_token( $begin_token, $end_token,1000)) {

   // upload $x into new $dest_cf which works great with no copy

   

    $begin_token = $cf->pack_key($last_key)

}

```

I've tried passing $last_key without passing it through pack_key to no avail.  I seems like there should be a way to get the token from a key - it's just using the correct hashing scheme right? 

Is it possible to iterate through a 5 billion row table, 1000 rows at a time to achieve this?    

Please say yes!  And if so, what am I doing wrong?

Cheers,

Chris