topic Re: Remote Temp and Voltage w/o Insight agents? in Server Management - Remote Server Management

Remote Temp and Voltage w/o Insight agents?

Vernon Fuller — Mon, 14 Nov 2005 12:27:04 GMT

Is it possible to get Temp and Voltage data from an iLO without Insight agents? Does the iLO have remote sensor capability, via IPMI 1.5 or a propritary interface? Does the motherboard ?

I am retro fitting a large cluster of proliant's with basic management features. I'm not supposed to put Insight agents on the OS, if at all possible. I have ML530, DL380, DL385, DL580, DL585. Current Ilo version is 1.64 but I will upgrade. Thank you for considering this.

Re: Remote Temp and Voltage w/o Insight agents?

David Claypool — Mon, 14 Nov 2005 14:22:12 GMT

Raw data on temperature and voltage is meaningless. ProLiant servers are self-managing as long as you use the ProLiant Support Pack.

Re: Remote Temp and Voltage w/o Insight agents?

Vernon Fuller — Mon, 14 Nov 2005 17:41:01 GMT

I can't install ProLiant Support Pack agents on the managed server, only on the one server that monitors the clustor. Super computer folk are *very* sensitive about extra applications. I can load drivers and libraries but no daemons, services or applications. This is why I was hoping there would be an out-of-band way to get at the data. Since the iLO card cannot support it, is there a BMC on the motherboard that talks IPMI 1.5 or a propritary interface?

Re: Remote Temp and Voltage w/o Insight agents?

Vernon Fuller — Wed, 16 Nov 2005 12:10:25 GMT

From further conversations:
HP has never adopted IPMI on ProLiant 300/500/700 because of security
concerns and a lack of standardization among parts manufacturers.
ProLiant manageability sets the standard for the industry. The Advanced
Systems Management Controller that is on the die with iLO monitors all
aspects of system health. You can choose the behavior of the system
through the Automatic Server Recovery feature based on an error
condition due to something like an overtemp: either continue, reboot or
shutdown. That action will be logged when it happens in the Integrated
Management Log, a copy of which is kept onboard iLO and is accessible
even if the system is not available.

ProLiant Servers have multiple thermal sensors which are individually
calibrated to the specs of the components they protect. In order to
know if there is an overtemp condition, you have to know the current
reading plus the threshold item--for each of the sensors. These
thresholds are different from model to model and even from configuration
to configuration. That's why the ASM controller takes care of that.
Knowing the actual temperature isn't useful because you cannot infer
ambient temperature from the sensors, so the data is only meaningful in
relation to the thresholds. There's a similar case when it comes to
voltage.

Use of the ProLiant Support Pack is a warranty requirement. Since you
say it's an HPC application, does that mean that the OS is Linux? If
so, and if the Insight agents are not an option for you, you do have one
alternative, and that is to deploy the 'hpasm' health package without
the agents. That package surfaces the instrumentation of the ASM chip
and installs a little command-line utility called hplog. Among its many
options is 'hplog -t' which provides the current reading for all
installed components and current and threshold temps.