CloudLens for Elastic Block Storage (EBS) monitors and reports risk events when disk performance data exceeds certain thresholds or when snapshots for data backup are not created. Prompt event handling is crucial to maintain stable disk operation and ensure data security. This topic describes the event types CloudLens for EBS can report, provides detailed event descriptions, and offers handling recommendations for various events.
Event property description
Event property | Description |
EventName | The name of the event. |
ResourceId | The ID of the disk associated with the event. |
ResourceType | The type of the resource associated with the event. Example: disk. |
Description | The description of the event. |
EventType | Event type:
|
EventLevel | Event level:
|
EventId | The event ID, which is unique for each resource. |
EventStatus | Event status:
|
StartTime | The start time of the event. |
EndTime | The end time of the event. |
RecommendAction | The action recommended for the event. Valid values:
|
RecommendParam | Parameter used in processing the event. For example, this parameter is set to a recommended provisioned value when the system processes a Cost Optimization event. |
You can query the properties of specific events on the Risk Events page in the console or by using the DescribeEvents API operation.
Event handling suggestions
The table below provides suggestions for handling different types of events.
Event name | Description | Trigger frequency | Event type | Event level | Handling suggestions | Determine whether the event is recovered |
Data protection | You did not create snapshots for a disk for an extended period of time. In this case, you may be unable to restore the disk to the most recent state in the event of ransomware, which may cause data loss. | Every morning | Alert | Warn | After the disk is restored, a new event in the Recovered status is pushed for the disk the next morning. The historical Active events will not disappear. | |
Cost optimization |
You can change the provisioned values to balance performance and cost. For more information, see ESSD AutoPL disk. | Once a week | Alert | Info | After the disk is restored, no events of the same type are pushed. | |
Disk I/O Hang | An I/O hang occurred when the operating system became unstable or experienced downtime due to the excessively high read/write I/O latency of file systems on a disk. | Real time | SystemException | Critical | View the performance metrics of disks to identify the cause. For more information, see View disk monitoring information. For information about how to detect I/O hangs on Alibaba Cloud Linux operating systems, see Detect file system and block layer I/O hangs. | After the disk is restored, a new event in the Recovered status is pushed for the disk. The historical Active events will not disappear. |
Specification mismatch between the instance and disks | The total specifications of disks on an instance exceed the upper limit supported by the instance type. The maximum disk performance may be limited by the instance type. For example, if an ECS instance has a maximum IOPS of 60,000 and a disk that has a maximum IOPS of 100,000 is attached to the instance, the performance of the disk is limited by the instance type.
| From 12:00 to 15:00 every day | Alert | Warn | When the total performance of disks on an instance exceeds the performance upper limit supported by the instance type, issues such as slow data processing and long response latency may occur. We recommend that you change the instance type based on your business requirements. For more information, see Upgrade the specifications of subscription instances and Change the specifications of pay-as-you-go instances. | After the disk is restored, no events of the same type are pushed. |
The IOPS of your instance reaches the upper limit | The total number of IOPS of disks attached to an instance reached the upper limit for the instance. For example, the maximum IOPS of an ECS instance is 60,000 and two disks areattached to the instance. The IOPS of a disk is 10,000 and the IOPS of the other disk is 51,000. The total real-time IOPS of the disks on the instance reaches the maximum IOPS of the instance.
| Within 5 minutes after the event is triggered Note The time is accurate to 5 minutes, and latency is measured in minutes. | Notification | Warn | An event in the Recovered status is reported when the event is triggered to remind you of these issues with the current disk. This does not involve whether the event is recovered. | |
The BPS of your instance reaches the upper limit | The total bytes per second (BPS) of disks on your instance reached the upper limit for the instance. For example, the maximum BPS of an ECS instance is 150 MB/s, and two disks are attached to the instance. The BPS of a disk is 100 MB/s, and the BPS of the other disk is 60 MB/s. In this case, the total real-time BPS (160 MB/s) of the disks exceeds the maximum BPS for the instance.
| Notification | Warn | |||
IOPS of a disk reached maximum IOPS per instance | The IOPS of your disk reached the upper limit for the instance. For example, an ECS instance has a maximum IOPS of 60,000, several disks are attached to the instance, and one disk has the IOPS of 70,000. In this case, the real-time IOPS of the disks exceeds the maximum IOPS for the instance.
| Notification | Warn | |||
BPS of a disk reached maximum BPS per instance | The BPS of your disk reached the upper limit for the instance. For example, the maximum BPS of an ECS instance is 150 MB/s, several disks are attached to the instance, and the BPS of one disk is 160 MB/s. In this case, the real-time BPS of the disks exceeds the upper limit of the BPS for the instance.
| Notification | Warn | |||
IOPS of a disk reached maximum IOPS per disk | The IOPS of your disk reached the upper limit for the disk.
| Notification | Warn | Issues such as slow data processing and high response latency may occur on the disk. For performance metrics of various types of disks, see Block storage performance.
| ||
BPS of a disk reached maximum BPS per disk | The BPS of your disk reached the upper limit for the disk.
| Notification | Warn | |||
Non-4K-aligned read/write operation | Read/write operations on your disk are not 4K-aligned, which may affect disk I/O performance. Note If read/write operations in partitions on the disk are not 4K-aligned, the disk may perform a read-modify-write operation across two 4K sectors. As a result, a read/write operation may involve multiple I/O operations, which degrades the performance of the disk. | Reported at 12:00 every day | Notification | Info | For more information, see I/O performance optimization: Improve I/O performance by using 4K alignment | |
Slow I/O detected on your disk | Your disk has slow I/O that lasts for 1 second or longer. | Real time | Notification | Warn | Slow I/O may affect applications that depend on disk performance, such as causing website loading delays. We recommend that you check whether your business is affected at the earliest opportunity. | |
Disk I/O burst | Disk I/O Burst occurred on your disk, which may incur performance burst fees. For more information, see ESSD AutoPL disk. | Every 1 hour | Notification | Info | We recommend that you check whether burst I/O of the disk meets your business expectations. | |
Disk performance burst fee cap | Your disk experienced a performance burst, and the total burst I/O triggered the performance burst fee cap rule | Every 1 hour, the total burst I/O of the past hour is calculated to determine whether the fee cap rule is triggered | Notification | Info | Although the performance burst fee cap rule can help you save on burst fees, we recommend that you check whether burst I/O meets actual needs to avoid unexpected business demands for burst I/O. |