Skip to main content
Scale a PostgreSQL cluster
Last update:

Scale a PostgreSQL cluster

Cloud database cluster PostgreSQL can scale. For example:

  • increase vCPU and RAM to improve cluster performance;
  • change fixed node configuration to arbitrary configuration or arbitrary configuration to fixed configuration — disk size must necessarily be larger than in the previous configuration, the number of vCPUs and RAM can be reduced or increased.

Check out the list node configurations.

Configuration change process depends on the availability of replicas.

Principle of operation⁠

The process of changing the configuration depends on the availability of replicas:

  • if there are no replicas in the cluster — the master node will be stopped while the changes are applied and the cluster will be temporarily unavailable for reading and writing;
  • if there are replicas in the cluster — the cluster will be read and write accessible while the changes are being applied. The replicas will be stopped one by one and the configuration of each replica will be changed. The master node will then switch to one of the replicas and the configuration of the past master node will be changed.

The configuration change time depends on the amount of data in the cluster.

For your information

If you have previously changed DBMS settings, they may be reset after scaling. Read more about settings PostgreSQL when scaling the cluster.

Scale up the cluster⁠

  1. Scale the cluster when users are least active.
  2. To avoid stopping a master node in a cluster with no replicas, we first add a line. When scaling is complete, the replica can be deleted.
  3. In control panel go to Cloud platform → Databases.
  4. Open the cluster page → tab Settings.
  5. Click Scale up the cluster.
  6. Select another fixed or arbitrary configuration. You can only select a configuration with more disk space. The number of vCPU and RAM can be reduced or increased.
  7. Optional: increase the number of lines in the cluster.
  8. Click Save.
  9. After scaling clear the DNS cache.
  10. Verify that the cluster has scaled correctly.

Clear DNS cache⁠

  1. Open the CLI.

  2. Clear the DNS cache:

    sudo systemctl restart systemd-resolved