Hi,
The documentation for upgrade of postgres server says that if postgres server is not using network storage, we need to deploy a new standby postgres server just in case of rollback. Can I ask why do we need to do that? Is it because of snapshot on local storage is not good enough as the one on network storage?
Another question is that if we have already a standby postgres server in another datacenter, will we still need to deploy a new standby postgres node for the upgrade?
Thanks.
Kyi Soe
Hi @kyisoethinkokyi, welcome back to the community! We saw your question and wanted to let you know we’re keeping it on our radar. We’ll also invite others in the community to pitch in and share their thoughts.
Meanwhile, we’d love for you to join this week’s tech talk 
Topic: Apigee MCP Tool Authorization at Scale
When: Thursday, May 21, 2026 (4 PM CEST | 10 AM EDT)
hi @kyisoethinkokyi - yes, your assumption is correct, but I would say, it’s an extra precautionary step to reduce the risk of downtime in case the primary PosgreSQL DB upgrade rollback would be necessary when you don’t already have one in place (either in the same region, or remote). If you have one extra secondary replica, instead of performing a more time consuming rollback you can always upgrade the secondary to primary DB (using the published operational documented steps), destroy the secondary and install a fresh secondary one of the same version pre-failed-upgrade.
The fact that a secondary is local or remote doesn’t really change in principle, but you should be aware of the data transfer times / speeds in case of a multi-TB DB when you setup a fresh replica, as those steps involve using pg_dump for the schema and data export, plus a remote secure copy of its tarball.
Hope this helps !
