Configuring properly the driver for rolling restart clusters

[Jean Carlo]

During some rolling restart, we faced some application doing timeouts.

Some request are in local_one and others in local_quorum, as my understanding, what we need to do is to use a retry policy on timeout or unavailable replica. But there was a discussion about migrating everything to local_quorum to avoid these timeouts. I think migrating to quorum will not fix this problem.

Am I correct ?

[Erick Ramirez]

Switching to QUORUM will make it worse, not better. Which version of the driver are you using?