2 articles tagged with Upgrades.
Sixty-seven enhancements shipped on August 26, but only a handful change how you run clusters. Scale-to-zero autoscaling, rootless kubelet, and a GA metrics API are the ones to read.
pg_upgrade is fast but takes downtime; logical replication lets you cut over while the old DB still serves traffic. The runbook, the gotchas, and the post-cutover checklist.
