CODEWITHSUNDEEP
pythoncassandradatastax
Chameleon
Chameleon

Reputation: 10128

Strange Cassandra 3.9 behaviour no update on bool field in Python

I am doing query to update time and bool flag but it not works at all? I have no idea what to do with this problem or how to handle. I spend hours with debugger but I see that it not works.

Query:

update task
set start_time = toTimestamp(now()), started = true
where id = d6283a7e-5b5c-11e7-98da-00059a3c7a00;

Field started is sometimes True or False - but often False - no idea why?

Full code:

    @classmethod
    def _db_start_task(cls, task_id, parent_task_id, farm_settings):
        assert isinstance(task_id, uuid.UUID)
        assert isinstance(farm_settings, FarmSettings)

        logger = logging.getLogger(__name__)
        session = cls._get_session(farm_settings)
        query_string = '''\
update task
set start_time = toTimestamp(now()), started = true
where id = %(task_id)s;''' % {
'task_id': task_id
}
        logger.debug('Query string is: %s.' % query_string)
        update = SimpleStatement(query_string=query_string,
                                 consistency_level=ConsistencyLevel.QUORUM,
                                 serial_consistency_level=ConsistencyLevel.SERIAL)
        warnings.warn('Statement execution problems is not handled.', UserWarning)
        update_result = session.execute(update, trace=True)

        warnings.warn('Debug code to remove start.', UserWarning)
        select = SimpleStatement(query_string='select start_time, started from task where id = %(task_id)s'  
                                 % { 'task_id': task_id },
                                 consistency_level=ConsistencyLevel.QUORUM,
                                 serial_consistency_level=ConsistencyLevel.SERIAL)
        result_set = session.execute(select)
        if result_set[0].started == False:
            logger.critical('Undefined behavior.')
        warnings.warn('Debug code to remove end.', UserWarning)

        if parent_task_id is not None:
            cls._db_count_started_children_tasks(parent_task_id, farm_settings)

Upvotes: 0

Views: 27

Answers (1)

Adam Holmberg
Adam Holmberg

Reputation: 7365

That is unexpected behavior with those consistency settings.

One possibility would be if you're encountering Unavailable errors, and your Cluster is configured using a DowngradingConsistencyRetryPolicy or something like it.

Upvotes: 1

Related Questions