IPMI SEL Parsing

[Bills, Jason M]

On 3/26/2021 6:47 AM, Bruce Mitchell wrote:
> On 3/26/2021 01:04, rgrs wrote:
>> Hi Matt, Bruce,
>>
>> Yes, SELs for sensor events, threshold alarms.
>> When a sensor value is beyond the range of thresholds, we need to log 
>> the SEL.
>>
>> My understanding of OpenBMC implementation:
>> dbus-sensors/phosphor-hwmon logs d-bus errors (logging/entry).
>> Each entry object has a "callout", that is used by phosphor-ipmi-host 
>> to convert dbus-error logs to IPMI SEL (16 bytes).
Please note that there are two locations that SEL entries can be logged:
1. D-Bus
2. Journal/syslog

It looks like you are using D-Bus which I'm not familiar with.

>>
>>
>> The issue I am facing:
>> 1. SEL creation: when I change sensor thresholds, I don't see a log 
>> generated for that sensor.
>> (phosphor-sel-logger has the capability to monitor threshold events 
>> and log SEL. But phosphor-hwmon doesn't generate the signal that 
>> phosphor-sel-logger is looking for.)
There is another thread on the mailing list that is discussing this 
issue: Question of phosphor-sel-logger.  We can use that thread to 
resolve this signaling issue.

>>
>> 2. Printing SELs: when I manually add logging/entry using busctl, 
>> ipmitool prints "Undetermined Hardware Failure".
>> (logging/entry/<number> is not having callout association to sensor/FRU)
>>
>> Where/what should I configure so that SELs are generated and parsed 
>> properly?
>> Different pieces of IPMI functionality is scattered among various 
>> packages, I'm not able to stitch them together.
>>
>> Thanks,
>> Raj
>>
>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>> On Friday, 26 March 2021 03:59, Bruce Mitchell 
>> <bruce.mitchell at linux.vnet.ibm.com> wrote:
>>
>>> On 3/25/2021 15:16, Matt Spinler wrote:
>>>
>>>> On 3/25/2021 7:49 AM, rgrs wrote:
>>>>
>>>>> Hi All, "ipmitool sel list" always shows the records as "Undetermined
>>>>> system hardware failure". Logging/entry doesn't have the proper
>>>>> association. How do I add association for log entries and respective
>>>>> FRUs? Which package creates Logging/entry Hi All,
>>>>> "ipmitool sel list" always shows the records as "Undetermined system
>>>>> hardware failure".
>>>>> Logging/entry doesn't have the proper association. How do I add
>>>>> association for log entries and respective FRUs?
>>>>> Which package creates Logging/entry object on sensor events? I don't
>>>>> see any call to SelAdd in phosphor-hwmon.
>>>>
>>>> What are the SELs for - threshold alarms, or some other sort of 
>>>> failures?
>>>>
>>>>> Thanks,
>>>>> Raj
>>>
>>> My assumption is that not everyone in the OpenBMC community has
>>> extensive use with IPMI, thus this just to share information.
>>> Sorry to those who already know this information and have been
>>> using it for years.
>>>
>>> This is share background as to the history or background of the
>>> IPMI SEL; probably more information than you need. But I would
>>> rather share 10% too much than be short by 0.1%.
>>>
>>>  From Intelligent Platform Management Interface Specification Second
>>> Generation v2.0
>>> Document Revision 1.1 October 1, 2013
>>>
>>> 31. System Event Log (SEL) Commands:
>>>
>>> The System Event Log is a non-volatile repository for system events and
>>> certain system configuration information. The device that fields the
>>> commands to access the SEL is referred to as the System Event Log Device
>>> or SEL Device.
>>>
>>> Event Message information is normally written into the SEL after being
>>> received by the Event Receiver functionality in the Event Receiver 
>>> Device.
>>>
>>> The SEL Device commands are structured in such a way that the SEL Device
>>> could actually be separated from the Event Receiver Device. In which
>>> case it would be the responsibility of the Event Receiver Device to send
>>> the appropriate ‘Add SEL Entry’ message directly to the SEL Device, or
>>> to pass the equivalent request through an intermediary.
>>>
>>> SEL Entries have a unique ‘Record ID’ field. This field is used for
>>> retrieving log entries from the SEL. SEL reading can be done in a
>>> ‘random access’ manner. That is, SEL Entries can be read in any order
>>> assuming that the Record ID is known.
>>>
>>> SEL Record IDs 0000h and FFFFh are reserved for functional use and are
>>> not legal ID values. Record IDs are handles. They are not required to be
>>> sequential or consecutive. Applications should not assume that SEL
>>> Record IDs will follow any particular numeric ordering.
>>>
>>> SEL Records are kept as an ordered list. That is, appending and deleting
>>> individual entries does not change the access order of entries that
>>> precede or follow the point of addition or deletion.
>>>
>>> ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- 
>>>
>>>
>>> Bruce
>>
>>
> 
> Adding Jason to the TO list.  Any thoughts on this Jason?
>