Skip to content
Percona Documentation

Changing PostgreSQL Options

You may require a configuration change for your application. PostgreSQL allows customizing the database with configuration files. You can use a ConfigMap to provide the PostgreSQL configuration options specific to the following configuration files:

Configuration options may be applied in two ways:

Note

PostgreSQL cluster is managed by the Operator, and so there is no need to set custom configuration options in common usage scenarios. Also, changing certain options may cause PostgreSQL cluster malfunction. Do not customize configuration unless you know what you are doing!

Use the kubectl command to create the ConfigMap from external resources, for more information, see Configure a Pod to use a ConfigMap.

You can either create a PostgreSQL Cluster With Custom Configuration, or use ConfigMap to set options for the already existing cluster.

To create a cluster with custom options, you should first place these options in a postgres-ha.yaml file under specific bootstrap section, then use kubectl create configmap command with this file to create a ConfigMap, and finally put the ConfigMap name to pgPrimary.customconfig key in the deploy/cr.yaml configuration file.

In both cases, the postgres-ha.yaml file doesn’t fully overwrite PostgreSQL configuration files: options present in postgres-ha.yaml will be overwritten, while non-present options will be left intact.

Creating a cluster with custom options

For example, you can create a cluster with a custom max_connections option in a postgresql.conf configuration file using the following postgres-ha.yaml contents:

---
bootstrap:
  dcs:
    postgresql:
      parameters:
        max_connections: 30

Note

dsc.postgresql subsection means that option will be applied globally to postgresql.conf of all database servers.

You can create a ConfigMap from this file. The syntax for kubectl create configmap command is:

kubectl -n <namespace> create configmap <configmap-name> --from-file=postgres-ha.yaml

ConfigMap name should include your cluster name and a dash as a prefix (cluster1- by default).

The following example defines cluster1-custom-config as the ConfigMap name:

$ kubectl create -n pgo configmap cluster1-custom-config --from-file=postgres-ha.yaml

To view the created ConfigMap, use the following command:

$ kubectl describe configmaps cluster1-custom-config

Don’t forget to put the name of your ConfigMap to the deploy/cr.yaml configuration file:

spec:
  ...
  pgPrimary:
    ...
      customconfig: "cluster1-custom-config"

Now you can create the cluster following the regular installation instructions.

Modifying options for the existing cluster

If you need to update cluster’s configuration settings, you should modify settings in the <clusterName>-pgha-config ConfigMap.

Note

This ConfigMap contains <clusterName>-dcs-config configuration applied globally to postgresql.conf of all database servers, and local configurations for the PostgreSQL cluster database servers: <clusterName>-local-config for the current primary, <clusterName>-repl1-local-config for the first replica, and so on.

For example, let’s change the max_connections option in a globally applied postgresql.conf configuration file for the cluster named cluster1. Edit the cluster1-pgha-config ConfigMap with the following command:

$ kubectl edit -n pgo configmap cluster1-pgha-config

This will open the ConfigMap in a local text editor of your choice. Make sure to modify it as follows:

...
cluster1-dcs-config: |
  postgresql:
    parameters:
     ...
     max_connections: 50
     ...

Now restart the cluster to ensure the update took effect.

You can check if the changes are applied by querying the appropriate Pods of your cluster using the kubectl exec command with a specific Pod name.

First find out names of your Pods in a common way, using the kubectl get pods command:

$ kubectl get pods
Expected output 
NAME                                              READY   STATUS    RESTARTS   AGE
backrest-backup-cluster1-j275w                    0/1     Completed 0          10m
cluster1-85486d645f-gpxzb                         1/1     Running   0          10m
cluster1-backrest-shared-repo-6495464548-c8wvl    1/1     Running   0          10m
cluster1-pgbouncer-fc45869f7-s86rf                1/1     Running   0          10m
pgo-deploy-rhv6k                                  0/1     Completed 0          5m
postgres-operator-8646c68b57-z8m62                4/4     Running   1          5m

Now let’s check the cluster1-85486d645f-gpxzb Pod for the current max_connections value:

$ kubectl -n pgo exec -it cluster1-85486d645f-gpxzb -- psql -c 'show max_connections;'
Expected output 
max_connections
-----------------
50
(1 row)
Last update: 2024-10-29