Cassandra column family not found in pycassa

Question

I'm attempting to connect to Cassandra in order to do a bulk insert. However, when I attempt to connect, I get an error.

The code I'm using:

from pycassa import columnfamily
from pycassa import pool

cassandra_ips = ['<an ip addr>']
conpool = pool.ConnectionPool('my_keyspace', cassandra_ips)

colfam = columnfamily.ColumnFamily(conpool, 'my_table')

However this fails on that last line with:

pycassa.cassandra.ttypes.NotFoundException: NotFoundException(_message=None, why='Column family my_table not found.')

The column family definitely exists:

cqlsh> use my_keyspace
   ... ;
cqlsh:my_keyspace> desc tables;

my_table

cqlsh:my_keyspace>

And I don't think this is a simple typo on the table name, as I've check it a dozen times, but also because of this:

In [3]: sys_mgr = pycassa.system_manager.SystemManager(cassandra_ips[0])

In [4]: sys_mgr.get_keyspace_column_families('my_keyspace')
Out[4]: {}

Why is that {}?

If it matters:

• The table/column family was created using CQL.
• The table is currently empty.

• The table was roughly created using:

  CREATE TABLE my_table (
    user_id int,
    year_month int,
    t timestamp,
    <tons of other attributes>
    PRIMARY KEY ((user_id, year_month), t)
  ) WITH compaction =
      { 'class' : 'LeveledCompactionStrategy', 'sstable_size_in_mb' : 160 };
  

Answer 1

In order to access CQL3 databases via a thrift API, such as pycassa, the tables must be created using compact storage.

CREATE TABLE my_table (
...
) WITH COMPACT STORAGE;

With regards to the primary keys, from the docs:

Using the compact storage directive prevents you from defining more than one column that is not part of a compound primary key.

Currently you are using a composite partition key but enabling compact storage limits us to using a compound partition key. So you will not have to limit it to a single column, it just has to be part of the compound key. One final reference.