NVIDIA GPU metrics in CloudWatch Explorer for AutoScaling Group

[thawro]

Is it possible to view the GPU metrics in CloudWatch Dashboard explorer? I would like to plot GPU metrics across my AutoScaling Group. I set up the GPU monitoring on my EC2 instances according to: tutorial, but all I can plot is a single instance gpu metrics (as normal line plot) and not in an explorer for the whole ASG.

[zhihonl]

Hi @thawro, thanks for reaching out! The agent is responsible for delivering the metrics to CloudWatch console, for dashboard question I suggest reaching out to AWS support for more clarifications. If you are missing GPU metrics, feel free to let us know and we can provide the debug process

[jefchien]

It sounds like you want to have the metrics sent with the AutoScaling Group as a dimension so you can see them in aggregate. I'm not sure what the configuration in the tutorial you followed is, but you'll need to add the AutoScalingGroupName dimension to your append_dimensions and aggregation_dimensions. See https://docs.aws.amazon.com/AmazonCloudWatch/latest/monitoring/CloudWatch-Agent-Configuration-File-Details.html for more details.

{
  "metrics": {
    "aggregation_dimensions" : [["AutoScalingGroupName"]],
    "append_dimensions": {
      "AutoScalingGroupName": "${aws:AutoScalingGroupName}"
    },
    ...
  }
}

[sky333999]

Please re-open if you require further assistance.