Re: [Freeipmi-devel] ipmimonitoring-sensors.c discretereading workaround

[Florian]

Hey Al,

That's awesome!

Sure, I'm totally OK with a git repo. As long as I just need to do configure and make I can do it.

Thanks for looking into this so fast!

Florian

On Wed, Dec 5, 2018 at 8:26 PM Albert Chu <address@hidden> wrote:

Hey Florian,

Ok, I think I found a bug.  The "discrete reading" flag is passed to an
underlying library correctly, but the result is not stored in
libipmimonitoring correctly.  I don't remember how this workaround
works 100%, because I never had this motherboard.

Once I look into this, how would like to be able to test?  Will
pointing you to a github repo be ok?

For my personal notes for later: see
_digital_sensor_reading() and _specific_sensor_reading(), store sensor
reading result and possibly units and other things too.

Al

On Wed, 2018-12-05 at 10:59 +0100, Florian wrote:
> Hey Albert,
>
> The changeing of record_ids and record_ids_length showed the sensor 5
> (but no data) like this
> Record ID, Sensor Name, Sensor Number, Sensor Type, Sensor State,
> Sensor Reading, Sensor Units, Sensor Event/Reading Type Code, Sensor
> Event Bitmask, Sensor Event String
> 5, Power Meter, 5, Current, N/A, N/A, N/A, 9h, 2h, 'Device Enabled'.
>
> IPMI_MONITORING_FLAGS_DEBUG unfortunately did not add ANY more logs
> at all, so I tried with IPMI_MONITORING_FLAGS_DEBUG_IPMI_PACKETS
> which gave me more logs.
> Those are quite a lot of more lines, so I pasted it into my
> privatebin: https://privatebin.florianstroeger.com/?58387298f9b913c4#
> CkDCOMP5eFXOs6Whn+giWtidWNhMZGrkhc3tXYPaK9o=
> I also tried setting discrete_reading to 1 as the sensor 5 was
> visible with record_ids, but that did output the same results.
>
> Florian
>
> On Wed, Dec 5, 2018 at 12:07 AM Albert Chu <address@hidden> wrote:
> > Hey Florian,
> >
> > Lets try:
> >
> > unsigned int record_ids[] = {5,
> > 0};                                                                
> >                                       
> > unsigned int record_ids_length = 1;
> >
> > and
> >
> > unsigned int ipmimonitoring_init_flags =
> > IPMI_MONITORING_FLAGS_DEBUG;
> >
> > Al
> >
> > On Tue, 2018-12-04 at 20:16 +0100, Florian wrote:
> > > Oops, I'm sorry, Android-Mail apparently deleted the CC...
> > > 
> > > I compiled with ipmimonitoring_init_flags = 1 now and got debug
> > stuff
> > > this time.
> > > 
> > > The output is following...
> > > (ipmi_monitoring_sensor_reading.c,
> > > ipmi_monitoring_get_sensor_reading, 1061): record_type '0xC0' not
> > > supported
> > > (ipmi_monitoring_sensor_reading.c, _get_sensor_bitmask_type,
> > 752):
> > > event_reading_type_code '0x70' bitmask is OEM
> > > (ipmi_monitoring_sensor_reading.c, _get_sensor_bitmask_type,
> > 752):
> > > event_reading_type_code '0x71' bitmask is OEM
> > > (ipmi_monitoring_sensor_reading.c, _get_sensor_bitmask_type,
> > 752):
> > > event_reading_type_code '0x71' bitmask is OEM
> > > (ipmi_monitoring_sensor_reading.c,
> > > ipmi_monitoring_get_sensor_reading, 1061): record_type '0x8' not
> > > supported
> > > (ipmi_monitoring_sensor_reading.c,
> > > ipmi_monitoring_get_sensor_reading, 1061): record_type '0x8' not
> > > supported
> > > (ipmi_monitoring_sensor_reading.c, _get_sensor_reading, 356):
> > > ipmi_sensor_read: sensor reading unavailable
> > > (ipmi_monitoring_sensor_reading.c, _threshold_sensor_reading,
> > 571):
> > > cannot read sensor for record id '23'
> > > (ipmi_monitoring_sensor_reading.c, _get_sensor_reading, 356):
> > > ipmi_sensor_read: sensor reading unavailable
> > > (ipmi_monitoring_sensor_reading.c, _threshold_sensor_reading,
> > 571):
> > > cannot read sensor for record id '24'
> > > (ipmi_monitoring_sensor_reading.c, _get_sensor_reading, 356):
> > > ipmi_sensor_read: sensor reading unavailable
> > > (ipmi_monitoring_sensor_reading.c, _threshold_sensor_reading,
> > 571):
> > > cannot read sensor for record id '38'
> > > (ipmi_monitoring_sensor_reading.c, _get_sensor_reading, 356):
> > > ipmi_sensor_read: sensor reading unavailable
> > > (ipmi_monitoring_sensor_reading.c, _threshold_sensor_reading,
> > 571):
> > > cannot read sensor for record id '39'
> > > (ipmi_monitoring_sensor_reading.c, _get_sensor_reading, 356):
> > > ipmi_sensor_read: sensor reading unavailable
> > > (ipmi_monitoring_sensor_reading.c, _threshold_sensor_reading,
> > 571):
> > > cannot read sensor for record id '40'
> > > (ipmi_monitoring_sensor_reading.c, _get_sensor_reading, 356):
> > > ipmi_sensor_read: sensor reading unavailable
> > > (ipmi_monitoring_sensor_reading.c, _threshold_sensor_reading,
> > 571):
> > > cannot read sensor for record id '41'
> > > Record ID, Sensor Name, Sensor Number, Sensor Type, Sensor State,
> > > Sensor Reading, Sensor Units, Sensor Event/Reading Type Code,
> > Sensor
> > > Event Bitmask, Sensor Event String
> > > ipmi_monitoring_sensor_read_sensor_bitmask_strings: success
> > > 
> > > 
> > > I'm unfortunately don't know much about C. You mean I should
> > change
> > > in 'unsigned int record_ids[] = {0};' the 0 to a sensor-id,
> > right?
> > > Because I did this and that neither worked with putting 5 in for
> > the
> > > power meter nor with any other sensor.
> > > 
> > > Florian
> > > 
> > > 
> > > 
> > > On Tue, Dec 4, 2018 at 7:47 PM Albert Chu <address@hidden> wrote:
> > > > Didn't see your output, I guess you forgot to paste it in.
> > > > 
> > > > Hmmmm.  That means the main monitoring code is just returning
> > "0",
> > > > i.e.
> > > > 0 sensors read.  Could you try experimenting with the settings
> > and
> > > > setting "record_ids" to the record we're trying to figure out
> > (i
> > > > think
> > > > it was #5 in your prior post).  Lets just concentrate on that
> > > > specific
> > > > record and try to figure out what's going on.  Also set
> > > > ipmimonitoring_init_flags for more debugging and lets see what
> > gets
> > > > output.
> > > > 
> > > > BTW, could you please respond to the mailing list.
> > > > 
> > > > Al
> > > > 
> > > > On Tue, 2018-12-04 at 07:50 +0100, Florian Ströger wrote:
> > > > > Hello Albert, 
> > > > > 
> > > > > Thanks for the fast reply!
> > > > > I've set 'int ignore_non_interpretable_sensors' to 0, but
> > when I
> > > > run
> > > > > the binary now it just outputs the headers, like this:
> > > > >                                                
> > > > -- 
> > > > Albert Chu
> > > > address@hidden
> > > > Computer Scientist
> > > > High Performance Systems Division
> > > > Lawrence Livermore National Laboratory
> > > > 
> > > > 
> > > 
> > > 
> > -- 
> > Albert Chu
> > address@hidden
> > Computer Scientist
> > High Performance Systems Division
> > Lawrence Livermore National Laboratory
> >
> >
>
>
--
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory

--

Florian Ströger

Zur Rossschwemme 5

3452 Atzenbrugg

****************************