Cluster Upgrade

By default, you do not need to manually upgrade a GaussDB(DWS) cluster. To upgrade a cluster on the console, see Delivering a Cluster Upgrade Task on the Console.

GaussDB(DWS) will notify you of any cluster O&M operation by sending SMS messages. Exercise caution when performing operations on the cluster during the O&M period.

If a node needs to be replaced due to a hardware fault, the repairCluster event will be triggered. You can check the event progress by Subscribing to Event Notifications.

If the upgrade affects the current query requests or service running, contact technical support for emergency handling.

A cluster is charged by hour or in yearly/monthly mode as long as it is in the Available state, so you will not see any difference in the bills if a faulty node or system upgrade causes a short interruption, for example, 15 minutes. If such events cause major system interruption, which is a very rare case, you will not be charged for those downtime hours.

Upgrading a Cluster

By default, you do not need to care about GaussDB(DWS) cluster patching or upgrading because GaussDB(DWS) will handle version upgrade automatically. After GaussDB(DWS) is upgraded, it will automatically upgrade the cluster to the latest version. During the upgrade, the cluster will be restarted and cannot provide services for a short period of time.

After a cluster is upgraded to 8.1.3 or later, it enters the observation period. During this period, you can check service status and roll back to the earlier version if necessary.
Upgrading the cluster does not affect the original cluster data or specifications.

The following figure shows the cluster version.

Figure 1 Version description
Click to enlarge

Service patch upgrade: The last digit of cluster version X.X.X is changed. For example, the cluster is upgraded from 1.1.0 to 1.1.1.
- Duration: The whole process will take less than 10 minutes.
- Impact on services: During this period, if the source version is upgraded to 8.1.3 or later, online patching is supported. During the patch upgrade, you do not have to stop services, but the services will be intermittently interrupted for seconds. If the destination version is earlier than 8.1.3, services will be interrupted for 1 to 3 minutes. Therefore, you are advised to perform this operation during off-peak hours.
Service upgrade: The first two digits of cluster version X.X.X are changed. For example, the cluster is upgraded from 1.1.0 to 1.2.0.
- Duration: The whole process will take less than 30 minutes.
- Impact on services: Online upgrade is supported for update to 8.1.1 or later. During the upgrade, you are not required to stop services, but services are intermittently interrupted for seconds. You are advised to perform the upgrade during off-peak hours.

Delivering a Cluster Upgrade Task on the Console

Prerequisites

Cluster 8.1.1 and later versions allow users to deliver cluster upgrade operations on the console.

Procedure

Log in to the GaussDB(DWS) management console.
In the cluster list, click the name of a cluster.
In the navigation pane, choose Upgrade Management.
On the Upgrade Management page, select a version from the Target Version drop-down list.
Click Upgrade. Click OK in the displayed dialog box.
Check whether the cluster is successfully upgraded.
- If the cluster version is 8.1.3 or later, the cluster enters the service observation period after the upgrade is complete. If you have verified your services, click Submit on the Upgrade Management page to complete the cluster upgrade. If you find your cluster performance affected by the upgrade, you can click Rollback to roll back the upgrade.
  - In versions earlier than 8.1.3, there is neither rollback nor submission button after the upgrade is complete.
  - If you do not submit the upgraded version, there will be a wlm thread which occupies the system storage space and affects the performance.
- If the cluster upgrade fails, click Rollback to roll back to the original cluster version, or click Retry to deliver the upgrade again.