[
https://issues.apache.org/jira/browse/CASSANDRA-17248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17607695#comment-17607695
]
Alex Petrov commented on CASSANDRA-17248:
-----------------------------------------
[~chovatia.jayd...@gmail.com] I took a (very) short glance, and can make the
following suggestions.
* I think a change to throw from log is a good idea.
* The test should probably not live in UFTest, the file was touched in the
patch for unrelated reasons; you can either create a new file or put it with
previous prepared statements tests.
* The source of the problem was the mixed version path. Unfortunately we could
not add the mixed version fuzz test to 3.0 because of the outdated driver. I
would suggest to add tests that exercise the mixed version path (ie where one
of the versions is pre-17248, and the other one has your patch), since this is
precisely where the issue has occurred. In other words, one of the nodes should
have new prepared statement behaviour enabled, and the other one should not.
Also, regular CQL test with new/new behaviour is advisable.
* The matter is further complicated by the fact that now we actually do have a
released version (between 17248 and this patch) where we return a hash without
keyspace for non-qualified statements. We need to test an upgrade path for
three implementations: pre-17248, between 17248 and your patch. The simplest
way to test this is probably a mixed version fuzz test.
* I'm not sure why remove {{warnAboutUseWithPreparedStatements }}call.
Unfortunately I'll be travelling for the next several weeks and may have
limited capacity for further reviews.
> Fix Prepared Statements behaviours after 15252
> ----------------------------------------------
>
> Key: CASSANDRA-17248
> URL: https://issues.apache.org/jira/browse/CASSANDRA-17248
> Project: Cassandra
> Issue Type: Bug
> Components: Messaging/Client
> Reporter: Alex Petrov
> Assignee: Alex Petrov
> Priority: Normal
> Fix For: 3.0.26, 3.11.12, 4.0.2
>
>
> [CASSANDRA-15252] has fixed an important issue: unwanted hash changes when
> preparing fully qualified prepared statements which was causing
> cluster-killing re-prepare loops. However, the fix introduced a regression:
> non-qualified statements can get prepared against one keyspace but then
> executed on another under some circumstances. This patch reconciles all
> behaviours.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
