You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@cassandra.apache.org by Runtian Liu <cu...@gmail.com> on 2026/03/05 17:56:18 UTC

[REVIEW] CEP-48 Strict Materialized View Consistency -- Patch Ready for Review

Hi dev@,

The implementation of CEP-48 (Strict Materialized View Consistency) is
ready for review.

For the full design, please refer to the CEP-48 wiki page:
https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=355568071

*Review Request*

This change touches the Paxos V2 hot path and the MV update pipeline. I'd
especially appreciate reviewers with experience in Paxos V2 / LWT internals
or Materialized View mechanics.

*Key areas for review:*

*MV mutation timing in Paxos.java*: MV mutations are applied after a
successful propose but before commit. Could a failed or superseded commit
leave MV state inconsistent?

*PaxosRepair changes (PaxosRepair.java)*: Strict MV tables force
re-prepare/re-propose instead of directly completing in-progress proposals.
Any liveness concerns?

*Repropose path under contention (Paxos.java)*: When a proposal is
superseded, MV mutations are applied via a read-then-apply sequence. Is
this safe under concurrent Paxos rounds?

JIRA: https://issues.apache.org/jira/browse/CASSANDRA-20880
Patch: https://github.com/apache/cassandra/pull/4358

Thanks,
Runtian