Help Center/ ModelArts/ Model Evaluation/ Viewing Model Evaluation Monitoring Alarms in AOM
Updated on 2026-08-28 GMT+08:00

Viewing Model Evaluation Monitoring Alarms in AOM

An alarm is an automated mechanism that evaluates monitoring metrics and log data in real time based on preset rules and automatically triggers alarm events when anomalies occur. It transforms scattered monitoring data into clear, actionable events, serving as a critical bridge between data observation and O&M action, helping you transition from passively responding to failures to proactively managing risks.

AOM provides alarm monitoring capabilities. For details about specific features, see AOM Alarm Monitoring Overview.

Model Evaluation Alarms

When a task execution fails, the system triggers an alarm. You can view alarm information related to model evaluation on the Alarm List page in the AOM console. If a task is deleted, the corresponding alarm will also be automatically cleared; cleared alarm records can be viewed in Historical Alarms. For details about how to view alarms, see Checking AOM Alarms or Events.

Alarms and repair suggestions related to model evaluation are listed in the table below.

Table 1 Alarm information

Function

Triggering Scenario

Alarm Name

Alarm Severity

Alarm Source

Alarm Message

Metadata

Possible Cause

Suggestion

Model evaluation

A minor alarm is reported after an evaluation job fails.

EvaluationJobExecutionFailed

Minor

MAS

Evaluation job failed, jobName: , failedTaskCount:

job_name,project_id

Dataset extraction or access failed.

Check the dataset configuration and recreate the job. If the fault persists, contact Huawei Cloud O&M personnel.