Azure Database for PostgreSQL - 单一服务器中的计划内维护通知Planned maintenance notification in Azure Database for PostgreSQL - Single Server

了解如何针对 Azure Database for PostgreSQL 的计划内维护事件做准备。Learn how to prepare for planned maintenance events on your Azure Database for PostgreSQL.

什么是计划内维护?What is a planned maintenance?

Azure Database for PostgreSQL 服务执行基础硬件、OS 和数据库引擎的自动修补。Azure Database for PostgreSQL service performs automated patching of the underlying hardware, OS, and database engine. 补丁包括新的服务功能、安全性和软件更新。The patch includes new service features, security, and software updates. 对于 PostgreSQL 引擎,次要版本升级自动进行,作为修补周期的一部分包含在内。For PostgreSQL engine, minor version upgrades are automatic and included as part of the patching cycle. 无需任何用户操作或配置设置即可进行修补。There is no user action or configuration settings required for patching. 此补丁经过广泛测试,并使用安全部署实践进行推广。The patch is tested extensively and rolled out using safe deployment practices.

计划内维护是一个维护时段,这些服务更新会在该时段内部署到给定 Azure 区域中的服务器。A planned maintenance is a maintenance window when these service updates are deployed to servers in a given Azure region. 在计划内维护期间,会创建通知事件,通知客户何时在承载其服务器的 Azure 区域中部署服务更新。During planned maintenance, a notification event is created to inform customers when the service update is deployed in the Azure region hosting their servers. 两次计划内维护之间的最短持续时间为 30 天。Minimum duration between two planned maintenance is 30 days. 你会提前 72 小时收到下一个维护时段的通知。You receive a notification of the next maintenance window 72 hours in advance.

计划内维护 - 持续时间和客户影响Planned maintenance - duration and customer impact

给定 Azure 区域的计划内维护通常应在 15 小时内完成。A planned maintenance for a given Azure region is typically expected to complete within 15 hours. 此时间范围还包括在必要时执行回退计划的缓冲时间。This time-window also includes buffer time to execute a rollback plan if necessary. Azure Database for PostgreSQL 服务器在容器中运行,因此数据库服务器重启通常需要 60-120 秒才能完成,但你的服务器在此 15 小时的范围内何时会受到影响是无法确定的。Azure Database for PostgreSQL servers are running in containers so database server restarts typically take 60-120 seconds to complete but there is no deterministic way to know when within this 15 hours window your server will be impacted. 工程团队会认真监视包括每个服务器重启在内的整个计划内维护事件。The entire planned maintenance event including each server restarts is carefully monitored by the engineering team. 服务器故障转移时间取决于数据库恢复时间,如果在故障转移时服务器上有大量的事务活动,这可能会导致数据库需要更长的时间才能联机。The server failover time is dependent on database recovery, which can cause the database to come online longer if you have heavy transactional activity on the server at the time of failover. 若要避免重启时间延长,建议在计划内维护事件期间避免任何长时间运行的事务(大容量加载)。To avoid longer restart time, it is recommended to avoid any long running transactions (bulk loads) during planned maintenance events.

总之,虽然计划内维护事件运行 15 小时,但单个服务器的影响通常只持续 60 秒,具体取决于服务器上的事务活动。In summary, while the planned maintenance event runs for 15 hours, the individual server impact generally lasts 60 seconds depending on the transactional activity on the server. 一个通知会在计划内维护开始前的 72 个日历小时内发送,另一个通知将在给定区域正在维护时发送。A notification is sent 72 calendar hours before planned maintenance starts and another one while maintenance is in progress for a given region.

如何收到有关计划内维护的通知?How can I get notified of planned maintenance?

你可以利用计划内维护通知功能接收有关即将进行的计划内维护事件的警报。You can utilize the planned maintenance notifications feature to receive alerts for an upcoming planned maintenance event. 你会在事件发生前的 72 个日历小时收到即将进行维护的通知,在给定区域正在维护时收到另一个通知。You will receive the notification about the upcoming maintenance 72 calendar hours before the event and another one while maintenance is in-progress for a given region.

计划内维护通知Planned maintenance notification


计划内维护通知目前在所有区域中都以预览版形式提供Planned maintenance notifications are currently available in preview in all regions

“计划内维护通知”允许你接收有关 Azure Database for PostgreSQL 即将进行的计划内维护事件的警报。Planned maintenance notifications allow you to receive alerts for upcoming planned maintenance event to your Azure Database for PostgreSQL. 这些通知与服务运行状况计划内维护集成,允许你在同一位置查看你的订阅的所有计划内维护。These notifications are integrated with Service Health's planned maintenance and allow you to view all scheduled maintenance for your subscriptions in one place. 它还有助于将通知扩展到不同资源组的适当受众,因为你可能有不同的联系人负责不同的资源。It also helps to scale the notification to the right audiences for different resource groups, as you may have different contacts responsible for different resources. 你会在事件发生前的 72 个日历小时收到即将进行维护的通知。You will receive the notification about the upcoming maintenance 72 calendar hours before the event.

我们将尽一切努力为所有事件提供 计划内维护通知 72 小时通知。We will make every attempt to provide Planned maintenance notification 72 hours notice for all events. 但是,对于关键或安全修补程序,通知可能会在事件快要发生时更晚一点发送,或者会被忽略。However, in cases of critical or security patches, notifications might be sent closer to the event or be omitted.

你可以在 Azure 门户上检查计划内维护通知,也可以配置警报以接收通知。You can either check the planned maintenance notification on Azure portal or configure alerts to receive notification.

从 Azure 门户检查计划内维护通知Check planned maintenance notification from Azure portal

  1. Azure 门户中,选择“服务运行状况”。In the Azure portal, select Service Health.
  2. 选择“计划内维护”选项卡Select Planned Maintenance tab
  3. 选择要为其检查计划内维护通知的订阅、区域和服务。 Select Subscription, **Region, and Service for which you want to check the planned maintenance notification.

接收计划内维护通知To receive planned maintenance notification

  1. 门户中,选择“服务运行状况”。In the portal, select Service Health.
  2. 在“警报”部分中,选择“运行状况警报”。In the Alerts section, select Health alerts.
  3. 选择“+ 添加服务运行状况警报”,并填写字段。Select + Add service health alert and fill in the fields.
  4. 填写所需的字段。Fill out the required fields.
  5. 选择“事件类型”,然后选择“计划内维护”或“全选”Choose the Event type, select Planned maintenance or Select all
  6. 在“操作组”中,定义接收警报的方式(获取电子邮件、触发逻辑应用等)。In Action groups define how you would like to receive the alert (get an email, trigger a logic app etc.)
  7. 确保“创建后启用规则”设置为“是”。Ensure Enable rule upon creation is set to Yes.
  8. 选择“创建警报规则”以完成警报Select Create alert rule to complete your alert

有关如何创建服务运行状况警报的详细步骤,请参阅 创建有关服务通知的活动日志警报For detailed steps on how to create service health alerts, refer to Create activity log alerts on service notifications.

能否取消或推迟计划内维护?Can I cancel or postpone planned maintenance?

维护是使服务器保持安全、稳定和最新状态所必需的。Maintenance is needed to keep your server secure, stable, and up-to-date. 计划内维护事件无法取消或推迟。The planned maintenance event cannot be canceled or postponed. 将通知发送到给定的 Azure 区域后,不能对该区域中的任何单个服务器进行修补计划更改。Once the notification is sent to a given Azure region, the patching schedule changes cannot be made for any individual server in that region. 补丁针对整个区域同时推出。The patch is rolled out for entire region at once.

是否所有 Azure 区域都同时修补?Are all the Azure regions patched at the same time?

否,所有 Azure 区域都按部署时段修补。No, all the Azure regions are patched during the deployment wise window timings. 在给定的 Azure 区域中,部署时段通常是下午 5 点到第二天的上午 8 点(当地时间)。The deployment wise window generally stretches from 5 PM - 8 AM local time next day, in a given Azure region. 地域配对的 Azure 区域会在不同的日期进行修补。Geo-paired Azure regions are patched on different days. 为了实现数据库服务器的高可用性和业务连续性,我们建议使用跨区域只读副本For high availability and business continuity of database servers, leveraging cross region read replicas is recommended.

重试逻辑Retry logic

暂时性错误也称为暂时性故障,是一种可以自行解决的错误。A transient error, also known as a transient fault, is an error that will resolve itself. 在维护期间可能会发生暂时性错误Transient errors can occur during maintenance. 系统在 60 秒以内可自动解决其中的大部分事件。Most of these events are automatically mitigated by the system in less than 60 seconds. 应使用重试逻辑来处理暂时性错误。Transient errors should be handled using retry logic.

后续步骤Next steps