cassandra replace primary key value

cassandra-replace-primary-key-value.py

"""
# table schema
CREATE TABLE raw.disk_storage (
    machine_id text,
    date date,
    timestamp timestamp,
    available bigint,
    total bigint,
    used bigint,
    PRIMARY KEY ((machine_id, date), timestamp)
)

# dependency
pip install cassandra-driver # https://datastax.github.io/python-driver/
"""

from cassandra.cluster import Cluster
cluster = Cluster([ip1, ip2, ip3])
s=cluster.connect()
s.set_keyspace('ks')

c = dict()
c["m1"] = "cc28f302-2b93-4267-b8cf-2e77596174d3.sysadmin"
c["m2"] = "8c8d7874-1820-4985-b0db-8e2f95f35c7e.sysadmin"
c["m3"] = "c28b4ba8-5204-4fae-af3d-ee8b179eca5a.sysadmin"
c["m4"] = "ded48033-1d86-4cfa-bab0-5677b2d5d3f0.sysadmin"
c["m5"] = "bf370d24-44b7-4fd6-b1a0-7f2d1b14b068.sysadmin"
date = ['2017-01-13', '2017-01-14', '2017-01-15', '2017-01-16']

pre_insert = s.prepare("insert into disk_storage (machine_id, date, timestamp, available, total, used) values (?, ?, ?, ?, ?, ?)")
pre_delete = s.prepare("delete from disk_storage where machine_id=? and date=? and timestamp=?")
def replace_machine_id(row, updated_id):
    s.execute(pre_insert, (updated_id, row.date, row.timestamp, row.available, row.total, row.used), timeout=3600)
    s.execute(pre_delete, [row.machine_id, row.date, row.timestamp], timeout=3600)
    
pre_key = s.prepare("select * from disk_storage where machine_id=? and date=?")
def get_rows_by_key(machine_id, date):
    rows = s.execute(pre_key, [machine_id, date], timeout=3600)
    return rows
    
for k in c.keys():
    t = 0
    for d in date:
        rows = get_rows_by_key(k,d)
        for row in rows:
            replace_machine_id(row, c[k])
print("done")