Platform For AI:View EAS events in CloudMonitor

View EAS events

Use the console

To view EAS events in the CloudMonitor console, perform the following steps:

1. Log on to the Cloud Monitor console.

2. In the navigation pane, choose Event Center > System Event.

3. On the Event Monitoring tab, select PAI for Product, and then click Search to view EAS system events.

4. In the Actions column of the target event, click Details. The following is an example:

   Parameters

   Parameter	Description
   Product	The product code. For PAI, the code is learn.
   Name	The event name. For a list of supported event names, see the Event name column in Appendix: Supported EAS events.
   Level	The event level. Possible values: INFO: Informational. WARN: Warning. CRITICAL: Critical.
   RegionId	The ID of the region where the service is located. For example, cn-shanghai for China (Shanghai).
   ResourceId	The resource ID. For more information, see Policy description.
   InstanceName	The name of the service or pod instance.
   Time	The time the event occurred, in UNIX millisecond timestamp format.
   GroupId	The CloudMonitor application group that corresponds to the EAS service. No group is assigned by default.
   Content	The core content of the event. Service-level events: A string. Pod instance-level events: A JSON object. For more information, see Content parameters.

   Content parameters

   Parameter	Description
   serviceName	The name of the service to which the instance belongs.
   serviceId	The ID of the service to which the instance belongs.
   serviceGroup	The service group to which the instance belongs.
   resourceType	The type of resource group where the instance resides. Possible values: PublicResource: a public resource group. DedicatedResource: a dedicated resource group.
   instanceType	The instance type.
   cpu	The number of CPUs used by the instance.
   memory	The memory size of the instance, in MB.
   gpu	The number of GPUs used by the instance.
   gpuMemory	The GPU memory size of the instance, in GB.
   nvidiaName	The name of the GPU used by the instance.
   role	The service role of the instance. Possible values: Queue: queue service. DataLoader: offline service. Standard: standard service.
   isBurst	Indicates whether the instance is managed by Burstable. Possible values: false: Burstable does not manage the instance, which means auto scaling is disabled for its resource group. true: Burstable manages the instance, which means auto scaling is enabled for its resource group.
   isSpot	Indicates whether the instance is a preemptible instance. Possible values: false: The instance is not a preemptible instance. true: The instance is a preemptible instance.
   callerUid	The UID of the user who created the EAS service.
   timestamp	The last startup time of the container.
   restartCount	The number of times the instance has been restarted.
   exitCode	The exit code of the instance. Empty by default.
   reason	The reason for the event.
   message	The event message.

Use the API

You can also use the DescribeSystemEventAttribute API to view EAS events.

Create and enable alert rules

Use the console

1. Create an alert rule for a system event. Configure the following key parameters:

   • Product type: Select PAI.

   • Event Type: Select ServiceInstance or Service.

   • Event Level: Select one or more event levels based on your business requirements.

   • Event name: Select one or more event names to monitor. For descriptions, see the Appendix.

   • Keyword Filtering: Specify keywords to filter events based on their content field.

2. Use callbacks for system event alerts.

Use the API

You can also use the API to create and enable alert rules. For details, see Create an alert rule for an event and Enable an event alert rule.

FAQ

Does a service instance refer to an inference service or a pod instance?

The Service event type represents service-level events. The ServiceInstance event type represents pod instance-level events. In this context, a service instance is a pod instance.

Appendix: Supported EAS events

EAS defines the following service-level and pod instance-level events.