Thread: Can PAF be used to provide zero downtime while primary and backup servers are being patched?

Our IT dept needs to install a patch on both primary and backup servers for our Postgres Automatic Failover configured DB (version 9.6 on linux).  From the standpoint of the DB users, can a strategy be implemented such that they see zero downtime during this process as the 2 servers are taken down in sequence for their patch, switching the role of primary and backup as this happens ?

On Thu, Aug  6, 2020 at 10:01:21AM -0400, David Gauthier wrote:
> Our IT dept needs to install a patch on both primary and backup servers for our
> Postgres Automatic Failover configured DB (version 9.6 on linux).  From the
> standpoint of the DB users, can a strategy be implemented such that they see
> zero downtime during this process as the 2 servers are taken down in sequence
> for their patch, switching the role of primary and backup as this happens ?

Assuming this is a minor relase upgrade, you can do it by just bouncing
the servers.  You can do the switch-over/switch-back too, but you have
to manage session migration.  I wrote a blog about it:

    https://momjian.us/main/blogs/pgblog/2018.html#October_1_2018

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EnterpriseDB                             https://enterprisedb.com

  The usefulness of a cup is in its emptiness, Bruce Lee