postgres-operator icon indicating copy to clipboard operation
postgres-operator copied to clipboard
verified publisher icon CrunchyData

pgupgrade with no downtime

Open fara-tode opened this issue 1 year ago • 2 comments

Overview

please provide the option to perform pgupgrade on pgcluster with at least 2 replicas with no downtime.

Use Case

strict SLAs

Desired Behavior

perform an upgrade on secondary first, after upgrade ensure secondary is in sync with leader, switchover to upgraded instance and upgrade other instances.

Environment

  • Platform: Kubernetes

Additional Information

fara-tode avatar Aug 27 '24 14:08 fara-tode

Hey @fara-tode! Just wanted to let you know that we have a story in our backlog to add this capability.

dsessler7 avatar Oct 16 '24 22:10 dsessler7

Hi,

Is that technically its possible "zero down time" in cluster mode while PGDB is upgraded to major version let us say, we have 2 instance PGDB1 & PGDB2 in cluster

PGDB1 is upgrade running for 30 min & during that 30 min PGDB2 is allowed read & write

PGDB1 makes meta data changes as part of upgrade & PGDB2 make changes like Insert, update, delete, index /view/ partition create/drop etc.

now we need to sync from both the DB, How it's possible.? and when this will be in sync

Ramki33 avatar Nov 15 '24 06:11 Ramki33