Disk statistics reporting very high KOPS

[hngrytgr]

The disk stats being reported seem to be very very high for operations per second

This is a raided set of 7200 rpm disks 40,000 operations a second seems high

Is this a calculation error or zabbix doing something strange?

[hngrytgr]

I am digging into the discovery Items. They seem to be doing odd things I do not understand. I am not a zabbix expert.
for example the item prototype for Disk:{#DEVICENAME}:Read:Ops per second
This gets this value from
UserParameter=custom.vfs.dev.read.ops[*],awk '{print $$1}' /sys/class/block/$1/stat

this field is

Name units description

---

read I/Os requests number of read I/Os processed
read I/Os, write I/Os

These values increment when an I/O request completes.

it is an integer that grows over time

a preprocessing rule is defined for this item.

changes per second which returns Numeric Float so it can have decimal values
123.987
but the item value is set to Numeric Unsigned
Is this the problem?

[hngrytgr]

So according to Zabbix that is a correct usage for the changes per second using Unsinged stomps the decimal places according to the docs
The Bytes a second is correctly multiplying by 512,

I am at a loss why it is reporting these incredibly high numbers

[hngrytgr]

I am beginning to think it is just reporting strangely because it is in a huge I/0 wait issue right about then like 4%.
It looks like digging into it everything is done correctly and reporting properly