LAS provides you with operations and monitoring capabilities for Key features, offering monitoring dashboards for common monitoring metrics. In addition, LAS integrates with Volcano Engine Cloud Monitoring to provide alerting, event callbacks, and other monitoring capabilities. This document provides an overview of LAS's operations and monitoring capabilities.

The following section introduces the main monitoring metrics and dashboard access points provided by the Key features modules.

Operator monitoring #

LAS provides the "Operator Service" feature. On the Operator Service page, you can view details such as the number of calls for each operator and the trend of token consumption. For more information, see Operator services.

Resource management (queues) #

LAS supports different types of queues. The monitoring capabilities and access points for different types of queues vary slightly.

• When using EMR Serverless queue resources directly as dedicated queue resources for LAS, you can go to the EMR Serverless queue resources page to view the monitoring metrics for the corresponding queue resources. For details, see Resource and Job Monitoring .
• When using the legacy LAS compute queues, you can view core resource monitoring metrics such as CPU and memory in the queue details. You can also view the operation log details for the queue in the operation logs.
  

Dataset #

Datasets provide an operation log feature. You can view the operation details for the corresponding dataset on the operation log page of the dataset details.

Development machine #

After creating a development machine, you can view core monitoring metrics such as CPU and memory on the monitoring page in the development machine details. You can also view the operation log details for the development machine on the operation log page.

List of events supported for monitoring #

LAS is also integrated with the Volcano Engine Cloud Monitoring Event Center, supporting the monitoring and recording of multiple events. The supported LAS cloud product events include core events for datasets, development machines, resource management, online services, job management, and workflow features. For details, see Cloud Monitoring Event Query .

When a corresponding event is triggered, an event record will be generated in the Cloud Monitoring Event Center. You can subscribe to and view these events using the event consumption capabilities of Cloud Monitoring. For example, you can configure a webhook to receive callbacks for these events.

Practical application: Cloud Monitoring event callback #

The following example demonstrates how to consume monitoring events in Cloud Monitoring by using the callback of cloud product events.

Preparation: View event parameters #

1. You can view relevant event records in "Cloud Monitoring - Cloud Product Events", for example, view events related to jobs (tasks) and workflows.
   
2. In the event details, check the event type (type), job template or workflow template ID (taskId), and current instance ID (resourceid). These will all be used in the webhook callback.
   

Create a notification template #

1. In "Cloud Monitoring - Notification - Notification Content Template", create a template. This template determines which JSON parameters will be sent in the callback POST request.

2. Configure callback template parameters.
   

   • Select "Event" as the notification type.
   • Select "Custom Callback" as the notification channel.
   • Use the default "Custom Callback" template for the template content.

Configure callback rules #

1. Create a callback address in "Cloud Monitoring - Notification - Callback".
   
2. Select "General Callback Address". After entering the address, you can test the POST callback.
   
3. Create an event rule in "Cloud Monitoring - Event Center - Event Rules"
   
4. You can use data.taskId to filter the tasks of interest, set up alert callbacks, and bind the previously created callback address and content template.
   

Event callback example #

After creating the event rule, if a new event occurs, it will trigger a callback. The server will receive JSON content as shown below:

{
  "Type": "Event",
  "AccountName": "xxx",
  "AccountId": "123",
  "Source": "AI数据湖服务 作业管理",
  "Id": "xxx",
  "EventType": "lake_ai_service_task:TaskInstance:taskInstanceRunFailed",
  "Description": "作业实例运行失败",
  "HappenedAt": 1769584967,
  "Region": "华北2（北京）",
  "Details": {
    "data": {
      "accountId": "123",
      "message": "UserCodeException",
      "region": "cn-beijing",
      "taskId": "t-uuid"
    },
    "datacontenttype": "application/json; charset=utf-8",
    "id": "6979b947293353438410****",
    "resourceid": "ti-uuid",
    "source": "lake_ai_service_task",
    "specversion": "1.0",
    "subject": "trn:las:cn-beijing:123:lake_ai_service_task/ti-uuid",
    "time": "2026-01-28T15:22:47+08:00",
    "type": "lake_ai_service_task:TaskInstance:taskInstanceRunFailed",
    "volcaccountid": "123",
    "volceventbusname": "default",
    "volcpublishtime": "2026-01-28T15:22:47+08:00",
    "volcregion": "cn-beijing",
    "volcresourcename": ""
  },
  "ProjectName": "",
  "TagSet": [],
  "Rules": [
    {
      "RuleId": "123",
      "RuleName": "test",
      "Level": "notice",
      "RuleURL": "https://console.volcengine.com/observe/cloud-monitor/event/rules/detail?id=123"
    }
  ]
}

• EventType Event type
• Details.resourceid: the task instance ID to which the event belongs

After the server receives the above callback, you can use the OpenAPI interface (task management/workflow) provided by LAS, and then use the SDK to obtain task instance information or start a new task instance.