CODEWITHSUNDEEP
phpcassandra
Harshal Borse
Harshal Borse

Reputation: 99

Update query in Cassandra 2.3

This is My Table structure as follows

CREATE TABLE manage_files_log (
    account_sid uuid,
    file_type text,
    file_sid timeuuid,
    date_created timestamp,
    file_description text,
    file_name text,
    status int,
    url text,
    PRIMARY KEY ((account_sid, file_type), file_sid)
) WITH CLUSTERING ORDER BY (file_sid DESC)

in this table I want to update my records with query 

UPDATE manage_files_log SET url='$url' WHERE account_sid =e40daea7-b1ec-088a-fc23-26f67f2052b9 AND file_type ='json' AND file_sid=961883e0-208f-11e6-9c41-474a6606bc87;

but it is inserting new record instead of updating existing record.

please help Me

below is the example of data in table where I want to update the url column value

account_sid                          | file_type | file_sid                             | date_created             | file_description | file_name | status | url
--------------------------------------+-----------+-------------------    -------------------+--------------------------+------------------+--------    ---+--------+-----
 e40daea7-b1ec-088a-fc23-26f67f2052b9 |      json | e15e02f0-20ab-11e6-9c41-474a6606bc87 | 2016-05-22 00:00:00+0000 |      descripton  |  testUrl1 |      1 |

Upvotes: 0

Views: 249

Answers (1)

Citrullin
Citrullin

Reputation: 2321

Okay, there's one thing you have to know about cassandra: Update or insert don't exist. I know, you write, in your query, update or insert, but it's the same. It's called Upsert. You might think: Whaaat? But there's a reason for this: Cassandra is a masterless distributed database system. You have, generally, no transactions. If you insert a value to node1 and want to update it after 10ms on node2, it could happen that your first value isn't on node2. With a hard UPDATE, your second operation will fail. But cassandra ignores this fact and writes the values to node2. After a time node2 and node1 will synchronizing their values. At this stage node1 gets the right values from node1. (Cassandra uses an internal column timestamp for synchronizing)

But you can also use update as update: Simply add IF EXISTS; to your query. But remember one thing: It's a big performance killer. Cassandra has to read all values from all nodes!

Upvotes: 4

Related Questions