Monitoring Upgrade Guidance

Upgrade Guidance for PostgreSQL Operator Monitoring

Migration to Upstream Containers

The Crunchy PostgreSQL Monitoring infrastructure now uses upstream Prometheus and Grafana containers. By default the installers will deploy the monitoring infrastructure using images from Docker Hub but can easily be updated to point to a Red Hat certified container repository. The Red Hat certified image catalog can be found here and the Docker Hub images can be found at the following links:

These containers are configurable through Kubernetes ConfigMaps and the updated PostgreSQL Operator Monitoring installers. Once deployed Prometheus and Grafana will be populated with resource data from metrics-enabled PostgreSQL clusters.

New Monitoring Features

Alerting

The updated PostgreSQL Operator Monitoring Infrastructure supports deployment of Prometheus Alertmanager. This deployment uses upstream Prometheus Alertmanager images that can be installed and configured with the metrics installers and Kubernetes ConfigMaps.

Updated pgMonitor

Prometheus and Grafana have been updated to include a default configuration from pgMonitor that is tailored for container-based PostgreSQL deployments. This updated configuration will show container specific resource information from your metrics-enabled PostgreSQL clusters. By default the metrics infrastructure will include:

New Grafana dashboards tailored for container-based PostgreSQL deployments
Container specific operating system metrics
General PostgreSQL alerting rules.

Updated Monitoring Installer

The installer for the PostgreSQL Operating Monitoring infrastructure has been split out into a separate set of installers. With each installer (Ansible, a Kubectl job, or Helm) you will be able to apply custom configurations through Kubernetes ConfigMaps. This includes:

Custom Grafana dashboards and datasources
Custom Prometheus scrape configuration
Custom Prometheus alerting rules
Custom Alertmanager notification configuration

Updating from Pre-4.5.0 Monitoring

Ensure that you have a copy of any install or custom configurations you have applied to your previous metrics install.

You can upgrade the Grafana and Prometheus deployments in place by using the new installers. After you have updated the PostgreSQL Operator and configured the values.yaml, run the metrics update. This will replace the old deployments while keeping your pvcs in place.

To make use of the updated exporter queries you must update the PostgreSQL Operator and upgrade your cluster.