disk usage metrics is incorrect in case of multiple partitions/storage devices attached to system

[abhishek210394]

I am an AWS user and I have multiple EBS volumes attached to an EC2 instance. Disk usage metrics takes into account only 1 volume/block storage in graph.

Another interesting thing that I noticed is when I set up alerts on disk usage (like "Check if metric system.disk.used>10.00 GB"), the agent calculates the total disk usage across all the volumes/disks unlike graphs.

For eg. I have 2 EBS volumes attached: /dev/xvda1 of 20GB and /dev/xvdaf of 50GB. I see metrics of only /dev/xvda1 in graph but when I set up an alert like "Check if metric system.disk.used>10.00 GB", I get an alert where disk usage is sum of disk usage of /dev/xvda1 and disk usage of /dev/xvdaf which is inconsistent with metric shown in graph.

[dedm]

Thanks, @abhishek-plackal, we have plans to add the ability to choose the disk/volume for alerting.