MONITOR PIPELINE ACTIVITY

Question

Hi,

I'm working on monitoring the success or failure status of AML jobs on Log Analytics. However, I don't know which tables in Log Analytics contain logs of AML jobs. I followed this documentation (https://learn.microsoft.com/ms-my/azure/azure-monitor/reference/tables/amldatastoreevent) to find out the required table but I still do not know which table is correct.

Could someone help me to find the correct table?

Many thanks!

Answer 1

The table you mentioned in the documentation (amldatastoreevent) is one of the tables related to AML, but there are other tables that you might find useful for tracking AML job statuses.

1. AMLJobEvent: This table contains events related to Azure Machine Learning jobs. You can query this table to get information about the status, errors, and completion of your AML jobs. Example query:

      
      AMLJobEvent
      
      | where JobStatus == "Failed" or JobStatus == "Completed"
      
      | project JobId, JobName, JobStatus, StartTime, EndTime, ErrorDetails
      
   

2. AMLComputeClusterEvent: This table contains events related to compute clusters used by AML jobs. It can provide insights into the state of your compute resources. Example query:

      
      AMLComputeClusterEvent
      
      | where ClusterState == "Failed" or ClusterState == "Succeeded"
      
      | project ClusterName, ClusterState, Timestamp, ErrorDetails
      
   

3. AMLRunEvent: This table logs detailed events about individual runs within an AML experiment. You can track the progress and outcome of each run. Example query:

      
      AMLRunEvent
      
      | where RunStatus == "Failed" or RunStatus == "Completed"
      
      | project RunId, ExperimentName, RunStatus, StartTime, EndTime, ErrorDetails
      
   

4. AzureDiagnostics: This table may also contain relevant logs if you have configured diagnostic settings for your AML workspace to send logs to Azure Monitor. Example query:

      
      AzureDiagnostics
      
      | where ResourceType == "Microsoft.MachineLearningServices/workspaces"
      
      | where OperationName == "RunStatusChanged"
      
      | project OperationName, Status, TimeGenerated, Resource