Re: Discrete Sensors in OpenBMC

[Patrick Williams <patrick@stwcx.xyz>]

[-- Attachment #1: Type: text/plain, Size: 1810 bytes --]

On Tue, Jan 19, 2021 at 03:22:41PM +0100, Wludzik, Jozef wrote:
> Does anyone use or implement discrete sensor in OpenBMC? Redfish defines 
> discrete triggers that use discrete values to detect that threshold is 
> crossed and trigger action should be commited. Redfish does not mention 
> about discrete sensor/values in other topics/schemes than Telemetry 
> (e.g. MetricDefinition, MetricReportDefinition and MetricReport). I am 
> asking about it because we are trying to find an example of discrete 
> sensor to proceed with DiscreteTriggers implementation (aka Discrete 
> thresholds) from Triggers scheme.
> 
> Ref.:
> 
> Redfish Telemetry White Paper - 
> https://www.dmtf.org/sites/default/files/standards/documents/DSP2051_1.0.0.pdf
> 
> Thanks,
> Jozef
> 

My impression is that IPMI called a lot of things "sensors" which were
not really hardware sensors and they often became a discrete sensor.  So
far we haven't had any real hardware sensors that are discrete.  All of
the Sensor definitions are in phosphor-dbus-interfaces[1].

There seems to be a bit of a throwback to IPMI terminology here though
with Redfish Telemetry.  For telemetry it could be very useful to keep
track of things which are indeed discrete (but not really a "sensor").
For example, log-in attempts might be something useful to insert into
telemetry but have a discrete component ("successful", "failed").  For
cases like this, we wouldn't have them under the Sensors namespace in
dbus but whereever is relevant to their domain.  I think you could find
a lot of data out in various namespaces which might be interesting to
place into telemetry but are not "sensors".

1. https://github.com/openbmc/phosphor-dbus-interfaces/tree/master/xyz/openbmc_project/Sensor

-- 
Patrick Williams

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]