System events are scheduled underlying O&M events or unexpected maintenance events that affect the running status of ECS instances. System events occur when ECS instances are restarted, stopped, or released due to actions such as update and maintenance, invalid operations, system failure, hardware or software failure, subscription expiration, or overdue payments.
Differences between routine maintenance and system events
ECS provides routine maintenance for underlying physical servers and forestall potential failures to improve the reliability, performance, and security of ECS instances. When a failure or threat is detected in a physical server, ECS uses hot migration to migrate ECS instances from the at-risk server to a healthy server and ensures that the instances continue to run properly. These operations are called routine maintenance. During routine maintenance, the system does not send notifications and the instances are not affected.
For system events, ECS will send you notifications that contain information such as solutions and event cycles. For scheduled system events, ECS will notify you of the impacts of the system events on your instances and the scheduled execution points in time. You can back up data in a timely manner and make preparations on the application layer before you handle system events. You can also query system events that were handled within the last week and obtain data for troubleshooting and analysis.
Limits
Phased-out instance families do not support the system event feature.
System event types
The following table describes the types, impacts, and recommended solutions of ECS system events.
| Impact | Event type | Parameter | Recommended solution |
|---|---|---|---|
| Instance restart | Instance restart due to scheduled system maintenance | SystemMaintenance.Reboot | Complete the following operations at an appropriate point in time during the user
operation window:
|
| Unexpected instance restart | Instance restart due to an unexpected system error | SystemFailure.Reboot | For more information, see Automatic recovery events of instances. |
| Instance restart due to an unexpected instance error | InstanceFailure.Reboot | If you receive an event notification during or after the instance restart, we recommend
that you perform the following operations:
|
|
| Instance redeployment | Instance redeployment due to scheduled system maintenance | SystemMaintenance.Redeploy | For more information, see Redeploy an instance equipped with local disks. |
| Instance redeployment due to a system error | SystemFailure.Redeploy | For more information, see Redeploy an instance equipped with local disks. | |
| Instance stop | Subscription instance: instance stop upon expiration | InstanceExpiration.Stop | For more information about how to renew a subscription instance, see Renewal overview. |
| Pay-as-you-go instance: instance stop due to overdue payments | AccountUnbalanced.Stop |
We recommend that you keep sufficient balance for your payment account to prevent instances from being stopped due to insufficient account balance. |
|
| Instance release | Subscription instance: instance release upon expiration | InstanceExpiration.Delete | For more information about how to renew a subscription instance, see Renewal overview. |
| Pay-as-you-go instance: instance release due to overdue payments | AccountUnbalanced.Delete |
We recommend that you keep sufficient balance for your payment account to prevent instances from being stopped due to insufficient account balance. |
|
| Instance release | Instance release due to a creation failure | SystemFailure.Delete | For more information, see Instance fails to be created. |
| Impacts on disk performance | Severe impacts on disk performance | Stalled | On the application layer, isolate read and write operations on the disk or temporarily remove the ECS instance from the backend server group of the SLB instance. |
| Damages to local disks | Damages to local disks | ErrorDetected | For more information, see Overview of system events on ECS instances equipped with local disks. |
| Instance restart and isolation of the damaged local disk | Instance restart due to scheduled system maintenance and replacement of the damaged local disk | SystemMaintenance.RebootAndIsolateErrorDisk | For more information, see Isolate damaged local disks. |
| Instance restart and restoration of the damaged local disk | Instance restart due to scheduled system maintenance and re-initialization of the damaged local disk | SystemMaintenance.RebootAndReInitErrorDisk | For more information, see Isolate damaged local disks. |
System event status
The following table describes the status of a system event during its lifecycle.
| Status | Status attribute | Description |
|---|---|---|
| Inquiring | Intermediate status | The system event is in the Inquiring state and waiting for your confirmation. The event enters the Executing state after it is confirmed. |
| Scheduled | Intermediate status | The system event has been scheduled but not performed. |
| Avoided | Stable status | The system event is responded to in advance within the user operation window. |
| Executing | Intermediate status | The system event is being executed. |
| Executed | Stable status | The system event has been executed. |
| Canceled | Stable status | The scheduled system event has been canceled. |
| Failed | Stable status | The system event failed. |
System event windows
System events have the following windows:
- User operation window: the period between the time when a system event is initiated and the time when the
system event is executed. Based on the impacts of the system event on your business,
you can select the recommended method to handle the event in advance or wait until
the default actions are triggered. If ECS fixes a system event triggered by a system
failure, ECS will send you an event notification in advance based on the system maintenance
schedule.
The following section describes the duration of the user operation window:
- For events in the Inquiring state, there is no time limit.
- For maintenance-related system events, the window is 1 to 30 days.
- For subscription instances that are to be stopped due to expiration, the window is three days.
- For pay-as-you-go instances that are to be stopped due to overdue payments, the window is less than one hour.
- Instances in which a system event occurs due to billing issues will be immediately stopped and released after 15 days if the issue is not resolved.
- Unexpected system events that are caused by failures or invalid operations do not have a user operation window.
- Event execution window: the period between the time when the system event is responded to and the time when
the event execution is completed. The system will send you the execution result.
The following section describes the duration of the event execution window:
- For system events such as failure recovery, the window is within 10 minutes.
- Unexpected system events that are caused by failures or invalid operations have only a short event execution window.
