Sorry if this has been around many times before, but I've done quite a bit of reading and searching for some type of human readable guide for setting up the lm_sensors for hardware monitoring. I've dorked around with the stuff in the system here, but after looking at the /etc/sensors.conf, I don't think I want to delve into that and try to pull out the right group of settings. It is definately a bit intimidating. So, is there any other software already written or in use for hardware monitoring that does not require a PhD to set up? The current sensors displays both the ISA and smbus readings, but none are even remotely close. The board uses thermistors for temp, which is mainly what I want to monitor.
sender: "Sam Drinkard" date: "Sun, Oct 09, 2005 at 08:58:07PM -0400" <<<EOQ
Sorry if this has been around many times before, but I've done quite a bit of reading and searching for some type of human readable guide for setting up the lm_sensors for hardware monitoring. I've dorked around with the stuff in the system here, but after looking at the /etc/sensors.conf, I don't think I want to delve into that and try to pull out the right group of settings. It is definately a bit intimidating. So, is there any other software already written or in use for hardware monitoring that does not require a PhD to set up? [..]
Sure:
up2date lm_sensors sensors-detect lots of Enter... ... chkconfig --level 2345 lm_sensors on service lm_sensors start sensors
That works for me, probably it should work for you too.
A good day to everyone, Alex
Hi Alex,
Thanks for the info. Unfortunately, the readings I get from sensors after those commands is nowhere even close to what some of the voltages should be. Core voltages are showing 4.08v and should be 1.3v. Lots of the others are way out as well, and temps are off, but higher than the actual hardware monitors in the bios show. I guess that is where all the different lines come into play in the sensors.conf file. There are just too many options to try to figure out which chip is doing what, especially with 3 or 4 different chips on the motherboard. I would have thought that some of the major mobo manufacturers would have written some kind of hardware monitor for linux, but guess not. Looks like I'll just have to dig into the /etc/sensors.conf and try to find the chip that works best, then set the highs and lows. Trouble with that is with the readings that are shown, nothing will be in spec. Any more suggestions before I screw things up ? :-)
Sam
Alexandru E. Ungur wrote:
Sure:
up2date lm_sensors sensors-detect lots of Enter... ... chkconfig --level 2345 lm_sensors on service lm_sensors start sensors
That works for me, probably it should work for you too.
A good day to everyone, Alex _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Thanks for the info. Unfortunately, the readings I get from sensors after those commands is nowhere even close to what some of the voltages should be. Core voltages are showing 4.08v and should be 1.3v. Lots of the others are way out as well, and temps are off, but higher than the actual hardware monitors in the bios show. I guess that is where all the different lines come into play in the sensors.conf file. There are just too many options to try to figure out which chip is doing what, especially with 3 or 4 different chips on the motherboard. I would have thought that some of the major mobo manufacturers would have written some kind of hardware monitor for linux, but guess not. Looks like I'll just have to dig into the /etc/sensors.conf and try to find the chip that works best, then set the highs and lows. Trouble with that is with the readings that are shown, nothing will be in spec. Any more suggestions before I screw things up ? :-)
The problem is that sometimes the same chips are used but different valued parts (resistors, etc.) are attached to the motherboard. So only the motherboard manufacturer has any real idea about the proper way to scale the value you can read from the chip to a concrete temperature or voltage. In many cases the default lm_sensors values work OK, but for some oddball motherboards with non-standard values (even though using standard chips!) there's nothing you can really do. You can either experiment or try to take a close look at the motherboard and figure out what actual elements are used (if that is even visible) - not the chips but the tiny extra elements like resistors and/or condensators... Of course you could also theoretically reverse engineer the windows drivers...
Cheers, MaZe.
Hee, hee, hee.. that sounds like a real undertaking to me :-) I guess with the hardware monitoring and protection the board itself provides (throttling cpu speeds) and such, I really shouldn't worry too much about temps. The reason I sent the Tyan board back was due to some overvoltage problems of cpu core voltage, and didn't want to wipe out a pair of Xeon cpu's. That is one area I really would like to monitor, but seeing what it shows now of 4.08 volts does not leave me much faith in things. Its just too much of a job to try to swag all the components on the mobo, as chip caps and resistors don't lend themselves to scrutiny by old eyes, even with a magnifier! Perhaps one day someone will get things figured out, but until then, I can take the readings with a grain of salt and watch for drastic changes from the initial.. that would be more useful probably in the long run
Thanks MaZe
Maciej Żenczykowski wrote:
The problem is that sometimes the same chips are used but different valued parts (resistors, etc.) are attached to the motherboard. So only the motherboard manufacturer has any real idea about the proper way to scale the value you can read from the chip to a concrete temperature or voltage. In many cases the default lm_sensors values work OK, but for some oddball motherboards with non-standard values (even though using standard chips!) there's nothing you can really do. You can either experiment or try to take a close look at the motherboard and figure out what actual elements are used (if that is even visible) - not the chips but the tiny extra elements like resistors and/or condensators... Of course you could also theoretically reverse engineer the windows drivers...
Cheers, MaZe.
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Sorry if this has been around many times before, but I've done quite a bit of reading and searching for some type of human readable guide for setting up the lm_sensors for hardware monitoring. I've dorked around with the stuff in the system here, but after looking at the /etc/sensors.conf, I don't think I want to delve into that and try to pull out the right group of settings. It is definately a bit intimidating. So, is there any other software already written or in use for hardware monitoring that does not require a PhD to set up? The current sensors displays both the ISA and smbus readings, but none are even remotely close. The board uses thermistors for temp, which is mainly what I want to monitor.
sure do the sendors-detect stuff and then install snmp and configure that - I use cacti and there are lm_sensors templates for cacti floating around and you can get what you want out of there
I looked at the pages on that and it seems to be a terrible amount of work to set up. It's not that important I guess, or either I'm getting lazy... :-)
Tom Brown wrote:
sure do the sendors-detect stuff and then install snmp and configure that - I use cacti and there are lm_sensors templates for cacti floating around and you can get what you want out of there
-
Alexandru E. Ungur -
Maciej Żenczykowski -
Sam Drinkard -
Tom Brown