cassandra upgrades multi-DC in parallel

[Carl Mueller]

If there are multiple DCs in a cluster, is it safe to upgrade them in
parallel, with each DC doing a node-at-a-time?