[
https://issues.apache.org/jira/browse/CASSANDRA-8767?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14324681#comment-14324681
]
Tyler Hobbs edited comment on CASSANDRA-8767 at 2/17/15 7:03 PM:
-----------------------------------------------------------------
It looks like the root of the problem is that ColumnFamily serialization
doesn't include a {{reversed}} flag. When the query is executed locally, the
{{reversed}} flag is set on the CF holding the results. When it's executed
remotely and then deserialized, it's not. The {{reversed}} flag is then used
to alter the comparator for adding cells to the CF. This causes the comparator
to be incorrect for the remote-execution case, leading to the assertion error.
was (Author: thobbs):
It looks like the root of the problem is that ColumnFamily serialization
doesn't include a {{reversed}} flag. When the query is executed locally, the
{{reversed}} flag is set on the CF holding the results. When it's executed
remotely and then deserialized, it's not. The {{reversed}} flag is then used
to alter the comparator for adding cells to the CF. This causes the comparator
incorrect for the remote-execution case, leading to the assertion error.
> "Added column does not sort as the last column" when using new python driver
> ----------------------------------------------------------------------------
>
> Key: CASSANDRA-8767
> URL: https://issues.apache.org/jira/browse/CASSANDRA-8767
> Project: Cassandra
> Issue Type: Bug
> Components: Core, Drivers (now out of tree)
> Environment: Cassandra 2.0.10, python-driver 2.1.3
> Reporter: Russ Garrett
> Assignee: Tyler Hobbs
> Fix For: 2.0.13
>
> Attachments: 8767-debug-logging.txt, describe-table.txt,
> exception-with-logging.txt, exception.txt
>
>
> We've just upgraded one of our python apps from using the old cql library to
> the new python-driver. When running one particular query, it produces the
> attached assertion error in Cassandra. The query is:
> bq. SELECT buffer, id, type, json FROM events WHERE buffer = %(bid)s AND
> idkey = %(idkey)s ORDER BY id ASC
> Where buffer and idkey are integer primary keys, and id is the clustering key
> (ordered asc).
> This query, with identical parameters, does not cause this error using the
> old cql python library, or with the cqlsh client..
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
