cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ondřej Černoš (JIRA) <>
Subject [jira] [Commented] (CASSANDRA-6625) Batch containing delete and insert leads to inconsistent results
Date Mon, 27 Jan 2014 17:03:38 GMT


Ondřej Černoš commented on CASSANDRA-6625:

[~brandon.williams] Original operations don't set the timestamp explicitly. From user's point
of view the result of batching operations is unexpected and confusing. If you consider this
Not a Problem, so be it, you are the maintainer. But the documentation of batch should state
it explicitly.

Mixing client-generated timestamps with implicit timestamps is fragile and error prone. Batches
therefore provide only limited value and one has to be very careful about the nature of batched
operations; if these have the slightest chance of dealing with the same value, batches need
to stay out.

Sidenote: from my understanding of Cassandra internals I know that fixing this batch misfeature
would be very difficult. So please fix at least the documentation. By no means am I the last
user hitting this wall. Thank you.

> Batch containing delete and insert leads to inconsistent results
> ----------------------------------------------------------------
>                 Key: CASSANDRA-6625
>                 URL:
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>         Environment: C* 1.2.11
>            Reporter: Ondřej Černoš
>            Priority: Minor
>              Labels: cql3
> On a single node cluster (i.e. ./bin/cassandra -f on localhost) we ran into the following.
Let's consider empty keyspace with the following table:
> {noformat}
>     a varchar,
>     b varchar,
>     PRIMARY KEY (a, b)
> ) WITH comment='List of a related to b - widerow';
> {noformat}
> The table is empty.
> Now we issue the following batch:
> {noformat}
> DELETE FROM test WHERE a = 'a1' AND b = 'b1';
> INSERT INTO test (a, b) VALUES ('a1', 'b1');
> {noformat}
> When the batch successfully finishes, the table is empty.
> This is consequence of the fact tombstone wins if timestamps are the same. And they are,
because the operation is batched.
> I consider this a bug. Batching operations shouldn't change the semantics of batched

This message was sent by Atlassian JIRA

View raw message