Generate UUID for Cassandra in Python

Question

Heh, I'm using

cf.insert(uuid.uuid1().bytes_le, {'column1': 'val1'}) (pycassa)

to create a TimeUUID for Cassandra, but getting the error

InvalidRequestException: 
InvalidRequestException(why='UUIDs must be exactly 16 bytes')

It doesn't work with

uuid.uuid1()
uuid.uuid1().bytes
str(uuid.uuid1())

either.

What's the best way to create a valid TimeUUID to use with the CompareWith="TimeUUIDType" flag?

Thanks,
Henrik

Answer 1

You must ensure your column family schema accepts UUID as key. Your code will work with a column family created as (using cassandra-cli):

create column family MyColumnFamily
  with column_type = 'Standard'
  and comparator = 'AsciiType'
  and default_validation_class = 'BytesType'
  and key_validation_class = 'TimeUUIDType';

To add values to this CF:

import pycassa
pool = pycassa.ConnectionPool('Keyspace1')
cf = pycassa.ColumnFamily(pool, 'MyColumnFamily')
cf.insert(uuid.uuid1(), {'column1': 'val1'})

Answer 2

Looks like you are using the uuid as the row key and not the column name.

The 'compare_with: TimeUUIDType' attribute specifies that the column names will be compared with using the TimeUUIDType, i.e it tells Cassandra how to sort the columns for slicing operations

Have you considered using any of the high level python clients? E.g. Tradedgy, Lazy Boy, Telephus or Pycassa