[ https://issues.apache.org/jira/browse/CASSANDRA-19489?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17833407#comment-17833407 ]
Berenguer Blasi commented on CASSANDRA-19489: --------------------------------------------- I actually already half-reviewed your PR yesterday when I had mixed feelings about it being a blocker so: - +1 on doing 5.0 and trunk and leaving 4.1 aside after the investigation on what else needs porting - +1 on the PR - TestUpgrade_indev_4_0_x_To_indev_trunk is a legit failure. We recently fixed timeouts on dtests so the ones we get now are most probably _not_ timeouts but legit. Also you can see by the stacktrace it is legit _but_ unrelated to this ticket imo. Can you please confirm with a local run and opening a ticket for it? It runs ok for me. Also this test has 3 other tickets open for other reasons so this seems like a new way of failing and the only one not env related at face value. - +1 otherwise > Guardrail to warn clients about possible transient incorrect responses for > filtering queries against multiple mutable columns > ----------------------------------------------------------------------------------------------------------------------------- > > Key: CASSANDRA-19489 > URL: https://issues.apache.org/jira/browse/CASSANDRA-19489 > Project: Cassandra > Issue Type: Improvement > Components: Consistency/Coordination, CQL/Semantics, Messaging/Client > Reporter: Caleb Rackliffe > Assignee: Caleb Rackliffe > Priority: Normal > Fix For: 5.0.x, 5.1 > > Attachments: ci_summary-1.html, ci_summary.html > > Time Spent: 1h 10m > Remaining Estimate: 0h > > Given we may not have time to fully resolve CASSANDRA-19007 before we release > 5.0, it would still be helpful to have, at the very minimum, a client warning > for cases where a user filters on two or more mutable (static or regular) > columns at consistency levels that require coordinator reconciliation. We may > also want the option to fail these queries outright, although that need not > be the default. > The only art involved in this is deciding what we want to say in the > warning/error message. It's probably reasonable to mention there that this > only happens when we have unrepaired data. It's also worth noting that SAI > queries are no longer vulnerable to this after the resolution of > CASSANDRA-19018. -- 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