How to query and configure scheduled events - ApsaraDB for HBase

ApsaraDB for HBase performs periodic operations and maintenance (O&M) activities -- such as minor version upgrades -- to keep your instances secure, reliable, and up to date. These activities are published as scheduled events that you can view, reschedule, or (in some cases) cancel from the console. In addition to the console, notifications are delivered through voice calls, emails, and internal messages so that you can plan ahead and minimize impact on your workloads.

Before you begin

Notification timing

Routine system maintenance events such as minor version upgrades: notifications are typically sent at least 3 days in advance.
Emergency risk-fixing events: notifications are typically sent 3 days or less in advance.

Configure notification recipients

Notifications are sent through voice calls, emails, internal messages, and the console. To make sure the right people are informed, log on to Message Center and turn on the switch for ApsaraDB Fault or Maintenance Notifications. We recommend that you add your database O&M engineers as message recipients.

Procedure

Log on to the ApsaraDB for HBase console.
In the left-side navigation pane, choose Event Center > Scheduled Event.
On the Scheduled Event page, review the event details.

The page displays the following fields: Instance ID, Region, Event Type, Cause, Business Impact, Scheduled Switching Time, and Status. For details about what causes each event type and how it affects your instance, see Event causes and impacts.
Perform one of the following operations on an event:
- Modify the Scheduled Switching Time
  You can set a custom scheduled switching time. The system determines when to perform the switchover based on the following priority: Scheduled Switchover Time > Default System Switchover Time.
  1. In the Scheduled Event list, select one or more events.
  2. In the upper-right corner of the Scheduled Event list, click Config Scheduled Event.
  3. In the Config Scheduled Event panel, set the scheduled switching time.
    
    To perform the switchover immediately, select Earliest Execution Time. After you confirm, the system begins preparing for the switchover right away and executes it as soon as preparations are complete.
    
    Note
    The Scheduled Switching Time cannot be later than the Latest Execution Time.
  4. Verify that the events selected for switching are correct.
    
    Click OK to confirm.
    
    If the selection is incorrect, return to the Scheduled Event list, select the correct events, and configure the scheduled switching time again.
- Cancel a scheduled event
  
  Currently, only Minor Version Upgrade scheduled events can be canceled. To cancel a Minor Version Upgrade event, select it and click Cancel Scheduled Event.

Event causes and impacts

Event cause

Impact type

Impact description

Minor version upgrade

Instance transient disconnection

After the scheduled switching time arrives, the switchover begins. See the deployment-specific details below.

Single-zone instance: One switchover occurs. The instance or the shards involved in the switchover experience a transient disconnection and enter a read-only state for up to 30 seconds while data is synchronized.
Multi-zone instance: Two switchovers occur. During each switchover, the instance or the shards involved experience a transient disconnection and enter a read-only state for up to 30 seconds while data is synchronized. The second switchover moves the primary node back to the primary zone to prevent issues such as increased access latency that can occur when the primary node remains in the secondary zone.

Differences between minor versions

The content updated in each minor version (kernel version) varies. For more information, see Release notes for minor versions.

Tip -- Best practices for scheduled switchovers

Schedule during off-peak hours. A switchover causes a brief transient disconnection. Performing it when traffic is lowest reduces the blast radius.
Implement a reconnection mechanism in your application. Your client should automatically reconnect after a transient disconnection so that the switchover is transparent to end users.
Understand multi-zone behavior. Multi-zone instances undergo two switchovers instead of one. Factor the additional read-only window into your maintenance planning.