Hardware Monitoring (temperature etc.) ?

lsf

I don't think it would hurt if we added this to the kernel conf, but I don't think we want to support it.
As this has a tendency to be either working or half working, depending on the HW it runs on.
Atleast i don't trust it to be correct as i have seen a lot of different results on all of the console montioring apps around.
In my experience only windows seems to get this mostly right, and still there are issues there as well.

Phobia

Well, I don't have the kind of experience that you folks do, but I tried all of the above mentioned hardware monitoring utilities from the shell, and all reported basically the same information. (except for -12v and -5v)

This kind of functionality "built in" would be terrific in my opinion, and even better if it were possible to be displayed in the webgui, perhaps on the system page that displays CPU usage? (I'm just talking about temperatures … I don't particularily need voltage info, though it is nice to be able to monitor all of that sort of information from the shell.)

I don't know if I'm typical, or just that 1% ... I'm not sure how many people use dedicated/embedded type hardware, or more like what I've done by recycling an old PC. All I can say is that as my setup is an older PC, this information is of great use as I'm trying to keep things as quiet as possible... I'd love to try passive cooling etc., but without the ability to monitor temperatures this would be a little difficult.

At any rate... I really appreciate all of the help that you've shown with getting these shell apps working... definately of great use to me!

-- Phob

Cyrandir

I have to admint, I agree with Phobia. I would like to have this kind of information at the ready, and having it integrated with the system status seems to make sense. Anyway, just another 1% to add… that makes 2% doesn't it?

hoba

I would agree adding it, maybe with an option to "enable display hardware senosrs" at the advanced page, so you can shut it down if it displays crap for your specific hardware. Other option would be to have this as package, so you can install/remove it when needed. Sending over the data to snmp would be a nice step as well, so you can trigger alerts when your cpu runs too hot for example because your cpufan died.

Phobia

Now that would be slick!!

What does one use to monitor SNMP? Are there tools available for Windows etc.?

It seems like the add-on package / option to enable sensors (likely with the default being off) would be a great idea. That way folks who didn't want to know or could care less wouldn't be bothered by said information. Would it be possible to add the information to the main status page with an add-on package? I noticed that other packages seem to be accessed through sub-menus and am unclear if this is required, or if added stuff from packages can reside in pre-exsisting pages in the webgui.

Great to talk about this feature… hopefully it will make the cut.

Thanks again,

-- Phob

hoba

If it's added as package it has to be a seperate page and a seperate menu entry. If it's in base with an option to turn it on/off it could be displayed at system>status.

sullrich

@hoba:

If it's added as package it has to be a seperate page and a seperate menu entry. If it's in base with an option to turn it on/off it could be displayed at system>status.

If someone wants to work on a package I'll be happy to integrate the kernel bits.

nima.m

I did some invenstigation and this is what I found, first we need to compile these parameters in Kernel :


	device          smbus
	device          ichsmb
	device          smb
	device          iicbus
	device          iicbb
	device          intpm
	device          pcf
	device          acpi

Then the best software out there that fit for pfsense is healthd
healthd is the only software that has support for so many chipset (W83781D, W83782D, W83783S, LM78, LM79 and the AS99127F)
this software is a deamon that easly can start in rc.d directory
later you can read out sensor figures by runing this command :
/usr/local/bin/healthdc -H
the -H parameters create a HTML output that can easly be included in pfsense WEBGUI

So the question is if Scott will compile a kernel with above parameters for pfsense :-)

sullrich

Those values are already present from what I can tell.

billm

@nima.m:

I did some invenstigation and this is what I found, first we need to compile these parameters in Kernel :
	device          smbus
	device          ichsmb
	device          smb
	device          iicbus
	device          iicbb
	device          intpm
	device          pcf
	device          acpi
Then the best software out there that fit for pfsense is healthd
healthd is the only software that has support for so many chipset (W83781D, W83782D, W83783S, LM78, LM79 and the AS99127F)
this software is a deamon that easly can start in rc.d directory
later you can read out sensor figures by runing this command :
/usr/local/bin/healthdc -H
the -H parameters create a HTML output that can easly be included in pfsense WEBGUI

Check out how the temp. monitor is setup on for 4801's. All the code for that is in /usr/local/www/* and subdirectories of it. Should be easy to modify to check for the presense of the healthd package and make the appropriate calls.

–Bill

KiaN

Works great for me with a cronjob :


*/1 * * * * /usr/local/bin/healthdc -H > /usr/local/www/temp/index.php

I only miss HD temp :(

# smartctl -a /dev/ad2s1a
smartctl version 5.33 [i386-portbld-freebsd6.1] Copyright (C) 2002-4 Bruce Allen
Home page is http://smartmontools.sourceforge.net/

Smartctl: Device Read Identity Failed (not an ATA/ATAPI device)

A mandatory SMART command failed: exiting. To continue, add one or more '-T permissive' options.

Phobia

Hi again all,

So, is this then possible to add to the actual install of PFSense, or is the manual hacking the only way to go?

Thanks,

– Phob

hoba

Unless somebody makes a package of this that generates a nice webguiscreen and/or maybe even some rrdgraphs "hacking" is the only option.

mbreitba

So are these values natively accessable via SNMP, or do I need to "hack" it to get those values? We already monitor a ton of stuff with MRTG (including bandwidth) but CPU usage and Memory usage would be nice to add to our graphs!

unforeseen

So I guess there was not enough interest for anyone to actually attempt this (release wise)? Seems like it would be "handy" if nothing else for the 3% of us… I would put in my two cents that if someone was to spend time to create it, I feel that it probably should be an add-on package. I'm really suprised though, up to the last post, it seemed like it was drawing some positive feedback.

All well, thanks for all the hard work!

ivanjong

i know this thread is old. but anyone else tried to run healthd and others on smbus ? i tried but i think our kernel doesnt have device smbus
device ichsmb
device smb
device iicbus
device iicbb
device intpm
device pcf
device acpi
is anyone sure this is compile in ? we have acpi. but not smbus and smb and iicbus and iicbb.