Top
Enterprise Postgres 17 for Kubernetes Overview

2.3 Configuration Change

Traditionally, changing FEP configurations such as postgresql.conf, pg_hba.conf, TLS certificates and keystore passphrase will require a redeployment of FEP server container. That causes an outage in a Highly Available environment.

A new CRD FEPConfig is defined to encapsulate those configurations. The operator will monitor the CR with this CRD definition and perform action accordingly to minimize outages. For example, operator will reload FEP daemon, instead of redeploying the FEP server container when a reloadable postgresql.conf parameter is changed. If a parameter change requires restart of FEP (e.g. max_connections), the operator will update the configuration file but defer the restart. End user can follow a defined procedure to restart the cluster manually at a scheduled maintenance time.