Question

I wrote a test program for testing Cassandra, and I had problems reading data. Seems like Cassandra sometimes takes one key for another.

Here is my test program :

from lazyboy import *
from lazyboy.key import Key
import uuid
import random

class TestItemKey(Key):
    def __init__(self, key=None):
     Key.__init__(self, "TestMX", "TestCF", key)

class TestItem(record.Record):
    def __init__(self, *args, **kwargs):
        record.Record.__init__(self, *args, **kwargs)
        self.key = TestItemKey(uuid.uuid1().bytes)

connection.add_pool('TestMX', ['localhost:9160'])

t1 = TestItem({'test':'foo'})
t1.key = TestItemKey(uuid.UUID('3cead15a-a54e-11df-87a2-000c298d2724').bytes)
t2 = TestItem({'test':'bar'})
t2.key = TestItemKey(uuid.UUID('3cebc15a-a54e-11df-87a2-000c298d2724').bytes)
t1.save()
t2.save()

print TestItem().load(t1.key.clone())
print TestItem().load(t2.key.clone())

(The chosen UUIDs are an example of the ones causing problems)

Here is the output of this script :

root@ubuntu:/mnt/hgfs/TestMX# python test.py 
TestItem: {'test': 'foo'}
TestItem: {'test': 'foo'}

Instead of the expected result :

root@ubuntu:/mnt/hgfs/TestMX# python test.py 
TestItem: {'test': 'foo'}
TestItem: {'test': 'bar'}

Note that the script usually works great with other randomely-chosen UUIDs, but sometimes not…

Answer 1

Sounds a lot like you’re hitting https://issues.apache.org/jira/browse/CASSANDRA-1235 which is fixed in the 0.6 branch and will be in 0.6.5, the next stable release.