Using other item values (Expression macros) in preprocessors

[tracyb_stp]

From what I can see, it is not possible to use another items value in a preprocessor.
Having this functionality would be useful for the following scenario.

Background:
When backups run on our servers, an agent runs as a low priority process, but uses as much CPU as it can. This results in ‘High CPU utilization’ alerts, which although true, do not indicate a performance issue, as the backup will be allocated less CPU if other processes need it.

Attempted solution:
New item:

Name: PercentProcessorTime For Below Normal priority processes
Key: wmi.getall[root\cimv2,"SELECT IDProcess,PercentProcessorTime FROM win32_PerfFormattedData_PerfProc_Process WHERE PriorityBase<8"]
Preprocessor 1: JSONPath: $.[?(@.IDProcess>"0")].PercentProcessorTime.sum()
Preprocessor 2: Custom multiplier: {? 1/ last(/Windows by Zabbix agent active/wmi.get[root/cimv2,"Select NumberOfLogicalProcessors from Win32_ComputerSystem"])}

New Trigger:

Name: High background process CPU usage in last 5 minutes
Expression: max(/Windows by Zabbix agent active/wmi.getall[root\cimv2,"SELECT IDProcess,PercentProcessorTime FROM win32_PerfFormattedData_PerfProc_Process WHERE PriorityBase<8"],5m) >50
Severity: Information

Updated trigger:

Name: High CPU utilization
Depends on: High background process CPU usage in last 5 minutes

Discussion on attempted solution:
Preprocessor 1 above will sum up all the values of the percentage processor time for all entries which aren’t the idle process nor the _total process. The sum can up to <number of cores>*100%.
To get a total CPU percentage, rather than per core percentage, we need to divide by the number of cores, which can be obtain from last(/Windows by Zabbix agent active/wmi.get[root/cimv2,"Select NumberOfLogicalProcessors from Win32_ComputerSystem"])

Preprocessor 2 is not accepted with the error 'Incorrect value for field "params": a numeric value is expected.'

Options considered:
I considered using a calculated item, with the formula:

last(/Windows by Zabbix agent active/wmi.getall[root\cimv2,"SELECT IDProcess,PercentProcessorTime FROM win32_PerfFormattedData_PerfProc_Process WHERE PriorityBase<8"]) / last(/Windows by Zabbix agent active/wmi.get[root/cimv2,"Select NumberOfLogicalProcessors from Win32_ComputerSystem"])

however, the calculated item will lag the measured item, as it runs on its own schedule.

I also considered a using dependant item. It doesn't have the lag issue, but has the same issue as a preprocessor on the the base item, as processing of dependent items is done by preprocessors.

Workaround used:
As a workaround I have not created preprocessor 2, and I have moved the division into the trigger:

( max(/Windows by Zabbix agent active/wmi.getall[root\cimv2,"SELECT IDProcess,PercentProcessorTime FROM win32_PerfFormattedData_PerfProc_Process WHERE PriorityBase<8"],5m) / last(/Windows by Zabbix agent active/wmi.get[root/cimv2,"Select NumberOfLogicalProcessors from Win32_ComputerSystem"])
) > 50

However this means that I when I graph the item ‘PercentProcessorTime For Below Normal priority processes’ it maxes out at <number of cores>*100%, rather than 100%, which is not useful when graphing alongside ‘CPU privileged time’, and ‘CPU utilization’.