Upgrade modes and the information required to upgrade Hologres instances in each mode - Hologres

This topic describes how to upgrade Hologres instances.

Upgrade modes

Supported upgrade modes

Hologres is backward compatible. Major version upgrades involve new features, whereas minor version updates involve issue fixes. We recommend that you upgrade your Hologres instances to the latest version.

Hologres instances can be upgraded in standard mode or hot upgrade mode. Both the standard mode and hot upgrade mode can be used for minor version updates and major version upgrades.

Note

Example of minor version updates: V1.3.25 to V1.3.36.
Example of major version upgrades: V1.1 to V1.3.
The upgrade duration differs based on the amount of data stored in Hologres instances. Service interruption caused by instance upgrades is not defined in the service level agreement (SLA).

Upgrade mode	Upgrade duration	Instance status	Impact on tasks	Description
Standard mode (services stopped)	5 to 10 minutes (The upgrade process includes data backup, version upgrade, and system initialization. The upgrade duration differs based on the data amount and metadata amount.)	Instances stop providing services when they are being upgraded.	Fully managed Flink tasks that involve data writes to Hologres tables must be stopped during instance upgrades and be started after the upgrades are complete. Otherwise, data may be lost. DataWorks data integration tasks and Blink tasks do not need to be stopped. During upgrades, a failover is triggered for the tasks and data is automatically restored based on the configured failover policy. We recommend that you set the number of failover retries to more than 10.	For general-purpose instances and virtual warehouse instances, you can manually upgrade them in the Hologres console. For more information, see Manual upgrade in this topic. For primary/secondary instances, you can join the Hologres DingTalk group to ask Hologres O&M engineers for upgrades in the background. This DingTalk group presents online technical support from Hologres experts. For more information about how to obtain technical support, see Obtain online support for Hologres. Note The service interruption duration for a standard upgrade varies based on the instance specifications and the amount of data stored in the instance. In special cases, the service interruption duration may be longer. If an instance is upgraded across multiple major versions, such as an upgrade from V0.10 to V1.3, the upgrade takes about 10 to 30 minutes. The specific upgrade duration varies based on the instance specifications and the amount of data stored in the instance. The upgrade does not cause changes in the endpoints of the instances, but may cause changes in the IP addresses of the instances.
Hot upgrade mode	10 to 30 minutes (The upgrade process includes data backup, transitional instance creation, transitional instance upgrade, traffic switchover, transitional instance release, and instance initialization. The upgrade duration differs based on the data amount and metadata amount.)	Instances are read-only for about 5 to 10 minutes during the upgrade. You can query data from the instances, but cannot write data to the instances or change metadata of tables in the instances.		Only instances of Hologres V1.1.82 and later can be upgraded in hot upgrade mode. You can join the Hologres DingTalk group to apply for a hot upgrade. This DingTalk group presents online technical support from Hologres experts. For more information about how to obtain technical support, see Obtain online support for Hologres. Note The upgrade does not cause changes in the endpoints of the instances, but may cause changes in the IP addresses of the instances. A hot upgrade across major versions is not supported. For example, a Hologres instance cannot be upgraded from V0.10 to V1.3 in the hot upgrade mode.

Upgrade modes supported by different types of products

The following table describes the upgrade modes supported by different types of products. You can select an upgrade mode based on your business requirements.

Product type	Standard mode (services stopped)	Hot upgrade mode
General-purpose instances and virtual warehouse instances	Supported. This upgrade mode is used by default. Note If you manually upgrade a general-purpose instance, the standard mode with services stopped is used by default.	Supported. This upgrade mode is optional. Note Hologres V1.1.82 and later support the hot upgrade mode. A hot upgrade can be performed only by Hologres O&M engineers.
Primary and secondary instances	Supported. This mode is used if you upgrade primary and secondary instances from V1.1 to V1.3.	Supported. This upgrade mode is optional. Note Hologres V1.3.62 and later support the hot upgrade mode. A hot upgrade can be performed only by Hologres O&M engineers. By default, the primary and secondary instances are upgraded together in the hot upgrade mode. During the upgrade, the primary instance is in the read-only state and is disconnected for a short period of time. You cannot write data to the primary instance that is in the read-only state. The secondary instances are also disconnected for a short period of time. This minimizes the impact on the primary and secondary instances.
Hologres Shared Cluster instances	Supported. This upgrade mode is used by default. Note Hologres O&M engineers are responsible for upgrades of Hologres Shared Cluster instances and will notify you of the upgrade time in advance in the product announcement. You are not allowed to separately upgrade a Hologres Shared Cluster instance.	Not supported.

Information required for upgrades

You need to ask Hologres O&M engineers to upgrade Hologres instances in the background in the following scenarios:

Upgrades of general-purpose instances
- The source instance version is earlier than V1.1.
  Note
  If the version of your Hologres instance is V1.1 or later, you can manually upgrade your Hologres instance in the Hologres console. For more information, see the Manual upgrade section in this topic.
- The hot upgrade mode is required.
Upgrades of primary and secondary instances

You need to provide related information if your Hologres instances need to be upgraded by Hologres O&M engineers in the background.

Minor version updates
Pre-upgrade checks of instances are not required for minor version updates. The instances can be directly updated. You can join the Hologres DingTalk group to apply for updates. You need to provide the IDs of your Hologres instances and the update time window. For more information about how to join the Hologres DingTalk group, see Obtain online support for Hologres.

Major version upgrades

Pre-upgrade checks of instances are required for major version upgrades. A pre-upgrade check takes about 1 to 3 days. You can join the Hologres DingTalk group to provide required information in the following table.

Note

To ensure compatibility between instance versions and retain the custom settings of parameters, a pre-upgrade check is required for the instance before the upgrade. The time window to perform the check is 1 to 3 days. Wait for the O&M engineers to inform you of the specific upgrade time.

Item	Description
Username	The name of your Alibaba Cloud account or a RAM user.
Instance ID	The ID of the Hologres instance that you want to upgrade. You can obtain the instance ID from the Basic Information section of the Instance Details page in the Hologres console. For more information, see Instance configurations.
Region and zone	The region and zone in which the Hologres instance that you want to upgrade resides. You can obtain the region and the zone from the Basic Information section of the Instance Details page in the Hologres console. For more information, see Instance configurations.
Instance version	The current version of the Hologres instance that you want to upgrade. You can obtain the instance version from the Basic Information section of the Instance Details page in the Hologres console. For more information, see Instance configurations.
Real-time write	If you use Blink, Realtime Compute for Apache Flink, or Apache Flink to write data in real time, provide the version number of the engine. Note If the version of the engine is outdated, data may be lost during the upgrade. To prevent data loss, make sure that you use Blink 3.7.9 or later, or Flink with the VVR version of 3.0 or later. If you use DataWorks to write data in real time, provide the real-time synchronization log and the region in which the resource group resides. If you use multiple exclusive resource groups for Data Integration, you must provide the real-time synchronization logs of all the resource groups.
Expected upgrade time	The point in time at which you want to start to upgrade the Hologres instance. Note The instance stops providing services when it is being upgraded. The duration of an upgrade varies based on the version and the specifications of the instance. The upgrade may last for a few minutes or even more than 10 minutes.

Manual upgrade

Hologres allows you to manually upgrade Hologres instances in the Hologres console. This allows you to upgrade your instances in a more convenient and efficient manner.

Limits

You can manually upgrade Hologres instances only in the standard mode. If you want to upgrade general-purpose instances in the hot upgrade mode, you need to contact Hologres O&M engineers for help.
You can manually upgrade Hologres instances of only V1.1 or later.
You cannot manually upgrade primary and secondary Hologres instances.
If you want to manually upgrade a Hologres instance to V2.2 or later as a RAM user, you must make sure that the AliyunServiceRoleForHologresIdentityMgmt role is assigned to the RAM user. For more information about the role, see Service-linked role for Hologres.
If you want to manually upgrade an instance as a RAM user, you must make sure that the AliyunHologresFullAccess policy is attached to the RAM user. For more information about the permissions of RAM users, see Grant permissions to a RAM user.
You must perform the upgrade within seven days after the pre-upgrade check is complete. Otherwise, you need to perform the pre-upgrade check again.

Procedure

A manual upgrade for a Hologres instance involves the following phases:

Note

We recommend that you manually upgrade Hologres instances during the off-peak hours of your business.

Click Start upgrade preparation to start a pre-upgrade check.
- The pre-upgrade check generally takes 1 to 3 hours.
- The instance works properly during the pre-upgrade check.
- The system performs operations such as compatibility checks and data backup in this phase.
Note
If the pre-upgrade check fails, contact the Hologres O&M engineers to obtain the cause of the failure and perform troubleshooting under the guidance of the Hologres O&M engineers.
After the pre-upgrade check is complete, click Upgrade now to upgrade the instance.
- The instance stops providing services when it is being upgraded.
- The upgrade may take 5 to 10 minutes. The specific upgrade duration varies based on the amount of data involved.
- If the upgrade is not complete in 30 minutes, the system performs a rollback operation by using the data that is backed up before the upgrade.

Example

This section describes how to manually upgrade a Hologres instance.

Log on to the Hologres console. In the left-side navigation pane, click Instances.
On the Instances page, click the name of the instance that you want to upgrade.
On the Instance Details page, click Version for upgrade next to the Version parameter.
Note
If the version of your instance is earlier than V2.2 and no service-linked role is created for your instance, a message appears indicating that you need to create the service-linked role AliyunServiceRoleForHologresIdentityMgmt when you upgrade your instance to V2.2 or later. After you confirm the authorization, you can proceed to the next step.
In the Upgrading an instance panel, check the version information and click Start upgrade preparation.
After the pre-upgrade check is complete, click Upgrade now.
Note
You can click Upgrade now to upgrade the instance within seven days after the pre-upgrade check is complete.
Your Hologres instance is upgraded to the latest version after the upgrade is complete.

Operation logs of manual upgrades

The events of the manual upgrade are recorded in ActionTrail. For more information about how to query event logs in the ActionTrail console, see Query event logs.

Solutions to common pre-upgrade check failures

Description	Solution
The upgrade fails.	By default, a timeout period of 30 minutes is configured for upgrades in standard mode. If an upgrade in standard mode is not complete within the timeout period, the system automatically rolls back the instance to the original version. The original data status is restored.
The following databases contain tables in the segment format. You cannot upgrade the instance to V2.0 or later. Update the data format of the tables. For more information, see Update the data storage format of existing column-oriented tables. Database: {Database}. Number of tables in the segment format: {TableUsingSegmentCount} Database: {Database}. Number of tables in the segment format: {TableUsingSegmentCount}	After you update the data format by following the instructions in Update the data storage format of existing column-oriented tables, click Terminate upgrade to terminate the upgrade. Then, submit the pre-upgrade check again to upgrade your instance to the latest version.
A Flink deployment to access binary logs in the HoloHub mode exists in the instance. You cannot upgrade the instance to V2.1 or later. For more information, see Default behavior changes.	After you perform operations by following the instructions in Default behavior changes and contact technical support, click Terminate upgrade to terminate the upgrade. Then, submit the pre-upgrade check again to upgrade your instance to the latest version.
A Flink deployment to access Hologres in the remote procedure call (RPC) mode exists in the instance. You cannot upgrade the instance to V2.0 or later. For more information, see Default behavior changes.	After you perform operations by following the instructions in Default behavior changes and contact technical support, click Terminate upgrade to terminate the upgrade. Then, submit the pre-upgrade check again to upgrade your instance to the latest version.