PageFaultsPerSecondStatistic is not taking care of MacOS

[DigitalBox98]

When starting the server on MacOS, the below error is raised :

[21:08:52,652] System.IO.DirectoryNotFoundException: Could not find a part of the path '/proc/stat'.

Looking at PageFaultStatistics.cs, it only handles Windows else it supposes it's Linux whereas it could be also MacOS :

public PageFaultsPerSecondStatistic()
{
if (RuntimeInformation.IsOSPlatform(OSPlatform.Windows))
{
performanceStatistic = new PerformanceCounterStatistic("Memory", "Pages/sec", null);
}
else
{
performanceStatistic = new LinuxPageFaultsPerSecondStatistic();
}
}

[NetDwarf]

Heh, yeah I pretty much ignored MacOS 😅. The abstraction was actually mostly to get rid of that error, while also giving some info. Of course MacOS was not a consideration, but it is if course possible to implement that as well.

What would be your plan for MacOS?

[DigitalBox98]

I know that the command on MacOS exists : sysctl -a

However for my information what is the aim of that code on other systems ?

[NetDwarf]

Sorry, forgot to answer, because I had no good one I guess 😅

I guess it should either give information or nothing.

[DigitalBox98]

Ok I've checked and now I understood.
A minimal version is to put dummy figures in order to avoid the errors when the statistic are displayed : CPU=0,0% DOL=0,0% pg/s=0,0 dsk/s=0,0

But it should be possible to have that kind of information :

for total CPU usage : top -l 1 | grep "CPU usage"
=> CPU usage: 2.30% user, 7.11% sys, 90.58% idle

for DOL usage : ps | grep DOLServer.dll | grep dotnet
=> 13156 ttys000 1:35.06 dotnet /Users/DOLSharp/Debug/DOLServer.dll
then top | grep 13156
=> PID COMMAND %CPU TIME #TH #WQ #PORT MEM PURG CMPRS PGRP PPID STATE BOOSTS %CPU_ME %CPU_OTHRS UID FAULTS COW MSGSENT
13156 dotnet 9.1 01:41.23 22 1 84 1232M 0B 0B 13152 13152 sleeping *0[1] 0.00000 0.00000 501 350119 793 57429 28695 5452334+ 287346+ 2683414+ 27 1313238+ 96.8 73964492 282739728 180 nicolas N/A N/A N/A N/A N/A N/A
(not sure which value from this one as 9.1% is not the accurate value)

for page faults : vm_stat | grep "faults"
=> "Translation faults": 127510514.

for disks : iostat -d
=> disk0
KB/t tps MB/s
24.40 15 0.36

Each type of information need to reprocess a little bit the output of the commands to retrieve the value

[NetDwarf]

Yeah with Linux this would be possible in a similar manner, but I didn't want to execute external commands, if I could avoid it. MacOS might not have that option (but I really don't know).

For the exact values I guess "translation faults" is something different. The 9.1% CPU value might actually be correct, if it is the utilization equivalent of a single thread i.e. if a 4 thread CPU can have 400% utilization total.

To be frank it would be totally okay, if you are not displaying some values for this OS or add others that seem to be of interest to you.