Cloudwatch Agent with nvme disks and setting up alerts

Question

When setting up disk metrics with the Cloudwatch agent and nvme disks, the nvme device is pushed to Cloudwatch as a dimension. The issue is that nvme devices can change after reboot (e.g. nvme0n1p1 can become nvme1n1p1). This causes issues when setting up Cloudwatch Alerts which will no longer work because the device dimension has changed between reboots potentially. Is there a way to not push the device to Cloudwatch or some other work around to prevent that nvme device from changing?

Accepted Answer

You probably can use "aggregation_dimensions" to list the dimensions you want to keep. You can refer to https://docs.aws.amazon.com/AmazonCloudWatch/latest/monitoring/CloudWatch-Agent-Configuration-File-Details.html#CloudWatch-Agent-Configuration-File-Metricssection for more info.

Answer

Have this exact same question. How to add mulitple mount points or drives.

Answer

Thanks for replying with the bits of your amazon-cloudwatch-agent.json required to get this to work.    
  
I was able to repeat what you did to create a disk_used_percent alert for my root volume (/), but can't seem to figure out how to create a disk_used_percent alerts for an additional attached volume (mounted to /mnt/data for example.)  Were you able to accomplish this?  If so, do you mind posting your amazon-cloudwatch-agent.json?  Thanks!

Answer

Awesome that worked! For anyone else who is curious, here is the magic that worked for me in my main metrics json object.

```
"aggregation_dimensions" : [["InstanceID"], ["InstanceId","fstype","path"]]
```

Cloudwatch Agent with nvme disks and setting up alerts

Relevant content