To provide a better experience and improve product performance and stability, Alibaba Cloud performs migration tasks from time to time. The system migrates some of your instances to new server nodes to upgrade software, hardware, and network facilities. This topic provides answers to commonly asked questions about instance migration.

Impact of instance migration events

Instance migration timeline
Time Description Impact
Alibaba Cloud generates a migration task Alibaba Cloud notifies you of the schedule of the migration task by using SMS, voice messages, emails, or internal messages. No impact.
Start time
Note The start time of a task is automatically generated based on the scheduled switchover time.
The system starts to perform operations related to instance migration, such as applying for new instance resources and synchronizing data.
  • You can use the database services, but you cannot perform instance-level operations, such as changing configuration and migrating across zones.
  • You cannot modify the scheduled switchover time when the system is performing the migration task.
Scheduled switchover time The system switches your workloads to the new instance.
  • The instance experiences transient connection errors and stays in the read-only state for 30 seconds or less to wait till all data is synchronized. Upgrade the instance during the off-peak hours and ensure that your application is configured with a reconnection mechanism.
    Note If the instance is of a cluster or read /write splitting architecture, each shard involved in a switchover experiences transient connection errors and stays in the read-only state for 30 seconds or less.
  • Data Management Service (DMS) and Data Transmission Service (DTS) are temporarily affected. After the switchover is completed, these services are restored to be normal.
Migration end time The task related to the instance is removed from the pending events. No impact. The zone, account, network, and endpoint of the instance are unaffected.

Start time and scheduled switchover time

  • Q: How long in advance will be a notification of a migration task sent?

    A: Due to the time gap from when the system sends a notification email to when the actual task is generated, the system sends a notification at least 40 hours in advance. After the scheduled time is reached, the system will switch over to the new instance during the upcoming maintenance window by default. You can adjust the scheduled switchover time on your own. For more information, see Query or manage pending events.

  • Q: How can I modify the scheduled switchover time?

    A: You can modify the scheduled switchover time in the ApsaraDB for Redis console or by using the related API operation. For more information, see Query or manage pending events and ModifyActiveOperationTask.

  • Q: Why do I fail to modify the scheduled switchover time?

    A: If the event has already been executed, (that is, the migration operation has already been started), you cannot change the scheduled switchover time.

  • Q: Can I estimate the end time of a migration task?

    A: Only the start time of the migration is displayed. The end time cannot be estimated because it is affected by multiple factors, such as the network latency, task queue, and data size.

Other FAQ

  • Q: Why are instances migrated?

    A: To provide a better experience and improve product performance and stability, Alibaba Cloud will perform migration tasks from time to time. The system migrates some of your instances to new server nodes to upgrade software, hardware, and network facilities.

  • Q: Will instances be migrated to other zones?

    A: No. It will only be migrated in the current zone.

  • Q: Can I cancel an instance migration task?

    A: You are not allowed to cancel the migration task because the instance migration is a task of high urgency. You can postpone a scheduled switchover time. For example, you can select a time during off-peak hours to perform the switchover task.

  • Q: Are data shards or proxy nodes in a cluster or read/write splitting instance switched over in parallel?
    A:
    • Data shards: The system performs the switchover in parallel.
    • Proxy nodes: The system performs the switchover in parallel. A maximum of half of proxy nodes in the ApsaraDB for Redis instance can be concurrently switched over.
  • Q: How will the system handle a failed switchover?

    A: The system confirms whether the execution is successful within one minute after a switchover task is completed. If the switchover fails, the system will re-initiate the switchover after a period of time.

  • Q: How can I confirm whether an instance migration task is completed?

    A: After a task is completed, the task related to the instance is removed from the pending events. You can create an event alert rule. Cloud Monitor monitors tasks and a notification is automatically sent to you when a task is completed. For more information, see Alert settings.