Can downtime be avoided when upgrading postgres database?


I tried searching but I couldn’t find any good resource on this.

We would like to upgrade the Postgres database (AWS Aurora) that Kong (v2.5.0) uses but we’d like to know if we can avoid downtime. Is it possible to do:

  1. Create a snapshot of the current database cluster
  2. Create a new and upgraded cluster based on the snapshot
  3. Point Kong to this new postgres cluster

There will be a significant amount of time between step #1 and step #2 and we’re worried that pointing the Kong cluster to a database behind by say a few hours would cause issues. Looking at the Postgres tables, it looks like there are Kong cluster events as well as rate-limiting information written in the tables.

Would doing the upgrade this way cause our Kong cluster to have issues? If not, what would be the likely effect to the Kong cluster? Loss of rate-limiting information, etc? I think we’re okay with rate-limiting hiccups, though.

Thank you in advance.