Using datastax-java-driver-core 4.13.0 to connect the Java application to Cassandra. When a Cassandra node goes down, all in-flight requests to that node fail with a DriverTimeoutException. This exception is not handled by DefaultRetryPolicy, meaning no automatic retry happens. The requests are non-idempotent, so implementing a custom retry mechanism risks duplicate writes. **Questions:** How can we gracefully handle DriverTimeoutException without causing duplicate writes? Is there a recommended way to retry only safe operations in such scenarios while preventing data inconsistency? Are there any DataStax-recommended strategies to mitigate request failures when nodes go down? Any guidance or best practices would be greatly appreciated!

Using DataStax Java driver version 4.17.0 against Cassandra version 4.1.4, a LOCAL_QUORUM lightweight-transaction write to the same DC as a single-DC keyspace shows a Write-SERIAL rather than a Write-LOCAL_SERIAL statistic. Is this expected behavior? I haven't dug into whether the LWT is accessing remote DCs; this would be undesirable behavior. Should it?