Re: [Freeipmi-devel] Recreating SDR cache before reading sensors information

[Al Chu]

You can pick specific sensors via the record-ID associated with the SDR
entry.

I'm not 100% sure what you're trying to program, but you may want to
consider looking at the libipmimonitoring library, which is a wrapper
that hides a lot of the sensor details.  There is an example on the
FreeIPMI webpage too.

Al

On Mon, 2012-02-27 at 01:50 -0800, Dinh Khac Thanh wrote:
> Hi,
> 
> BTW, this is unlikely, but is there any more efficient way to get
> sensor data for one particular sensor rather than getting all sensors'
> data?
> I just need confirmation. Thanks in advance.
> 
> Best regards
> Thanh
> 
> On Mon, Feb 27, 2012 at 1:49 PM, Dinh Khac Thanh <address@hidden>
> wrote:
>         Hi,
>         
>         Yes, I am using libfreeipmi instead of ipmi-sensors
>         executable.
>         
>         Best regards,
>         Thanh
>         
>         
>         On Mon, Feb 27, 2012 at 1:31 PM, Kaiwang Chen
>         <address@hidden> wrote:
>                 Thanh, are you using libfreeipmi instead of the
>                 ipmi-sensors
>                 executable? I am not quite sure about your senario,
>                 anyway good news
>                 is it works as you expect.
>                 
>                 Thanks,
>                 Kaiwang
>                 
>                 2012/2/27 Dinh Khac Thanh <address@hidden>:
>                 > Hi Kaiwang,
>                 >
>                 > I found out that It works if I reopen the cache
>                 before getting the sensor
>                 > data. Thanks anyway.
>                 >
>                 > Best regards
>                 > Thanh
>                 >
>                 >
>                 > On Mon, Feb 27, 2012 at 12:04 PM, Dinh Khac Thanh
>                 <address@hidden> wrote:
>                 >>
>                 >> Hi Kaiwang,
>                 >>
>                 >> Thanks for your reply. I tested with/without
>                 recreating the cache before
>                 >> getting the sensor data. It works fine when the
>                 cache is recreated, e.g:
>                 >> CPU1 Temp 46 C nominal
>                 >> CPU2 Temp N/A unavailable
>                 >> ...
>                 >> Fan11 N/A unavailable
>                 >> Intrusion N/A unavailable
>                 >> PS Status N/A unavailable
>                 >>
>                 >> but not when the cache is reused:
>                 >>
>                 >> PS Status N/A unavailable
>                 >> PS Status N/A unavailable
>                 >> PS Status N/A unavailable
>                 >> ...
>                 >> PS Status N/A unavailable
>                 >>
>                 >> The number of records in both cases are 30.
>                 >> The code is based on ipmi-sensors.c
>                 >>
>                 >> Best regards
>                 >> Thanh
>                 >>
>                 >>
>                 >> On Fri, Feb 24, 2012 at 9:12 PM, Kaiwang Chen
>                 <address@hidden>
>                 >> wrote:
>                 >>>
>                 >>> 2012/2/24 Dinh Khac Thanh <address@hidden>:
>                 >>> > Hi,
>                 >>> >
>                 >>> > I am working on a monitoring application that
>                 needs to get sensor
>                 >>> > information via IPMI every 10 seconds. I have
>                 skimmed through the
>                 >>> > source
>                 >>> > code for ipmi-sensors to learn how to use
>                 freeipmi.
>                 >>> >
>                 >>> > My question is: If I am using the same IPMI
>                 session to get sensor data,
>                 >>> > do I
>                 >>> > need to recreate the SDR cache every time before
>                 reading the data?
>                 >>>
>                 >>> I believe not. The cache is checked before real
>                 ipmi traffic, and it's
>                 >>> a noop when the cache exists and is up to date.
>                 There is one sdr cache
>                 >>> per ipmi node, which defaults  to
>                 >>>
>                 ~/.freeipmi/sdr-cache/sdr-cache-<localhost>.<ipminode>
>                 >>>
>                 >>> The logic resides in common code path defined in
>                 >>>
>                 
> common/toolsdr/tool-sdr-cache-common.c:sdr_cache_create_and_load(),
>                 >>> which is called by run_cmd_args(), which in turn
>                 called by
>                 >>> _ipmi_sensors(). You can find _ipmi_sensors as
>                 callback in main
>                 >>> function. All of them are in
>                 ipmi-sensors/ipmi-sensors.c, in the
>                 >>> trunk(revision 8603).
>                 >>>
>                 >>>
>                 >>> >
>                 >>> > Best regards
>                 >>> > Thanh
>                 >>> >
>                 >>> > _______________________________________________
>                 >>> > Freeipmi-devel mailing list
>                 >>> > address@hidden
>                 >>> >
>                 https://lists.gnu.org/mailman/listinfo/freeipmi-devel
>                 >>> >
>                 >>>
>                 >>> Thanks,
>                 >>> Kaiwang
>                 >>
>                 >>
>                 >>
>                 >>
>                 >> --
>                 >> Dinh Khac Thanh
>                 >> You can do anything, but not everything
>                 >> You miss 100 percent of the shots you never take
>                 >
>                 >
>                 >
>                 >
>                 > --
>                 > Dinh Khac Thanh
>                 > You can do anything, but not everything
>                 > You miss 100 percent of the shots you never take
>                 
>         
>         
>         
>         -- 
>         Dinh Khac Thanh
>         You can do anything, but not everything
>         You miss 100 percent of the shots you never take
>         
> 
> 
> 
> -- 
> Dinh Khac Thanh
> You can do anything, but not everything
> You miss 100 percent of the shots you never take
-- 
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory