add Node<x> as /redfish/v1/Systems/Node<x>/ for multi-node system

yugang.chen yugang.chen at linux.intel.com
Fri Jun 5 15:13:52 AEST 2020


Deepak,  It looks something like that. Is it possible to loop us in 
while you post something on Gerrit? thanks.

On 6/4/2020 4:14 PM, Deepak Kodihalli wrote:
> On 04/06/20 1:08 pm, yugang.chen wrote:
>> Dear Deepak,
>>
>> I'm openbmc engineer working for redfish' slave log in multiple 
>> systems, we want to add slave node's log to redfish,  is there some 
>> design or documentation for systems schema based on associations to 
>> different systems interface? if so, we can work based on that common 
>> interface for slave logs on redfish, thanks.
>
> This is something I'm still working on (and should be able to post 
> something on Gerrit in a few days). Just so that we're on the same 
> page - the problem I am trying to solve is a D-Bus model based 
> approach to depict multiple systems and the parts contained within 
> (the parts could be physical containment as well as logical things 
> like logs). This would prevent writing code that assumes specific 
> names in the D-Bus path hierarchy. Daniel/James - does this help the 
> problem you are trying to solve?
>
>> Best Regards
>>
>> Yugang(Daniel) Chen
>>
>>
>> On 6/3/2020 12:40 AM, James Feist wrote:
>>> On 6/1/2020 11:39 PM, yugang.chen wrote:
>>>> Dear All,
>>>>
>>>> May I have your attention? For a multi-node system,  there is an 
>>>> option for master to use name "Node<x>" as ​{ComputerSystemId} to 
>>>> stand for slave node, to show slave node's event log in master 
>>>> redfish,
>>>>
>>>> Could you please share your comment/idea for naming ("Node<x>")? 
>>>> thanks.
>>>
>>> Has there been any design docs for multiple systems in Redfish? This 
>>> is much bigger than just adding support for multiple logs, this is 
>>> adding support for multiple Systems interfaces, and many things need 
>>> to be considered such as the mapping of multiple CPUs, Memory, logs, 
>>> and how this fits into the inventory picture. On IRC a while back we 
>>> had discussed making the systems schema based on associations to 
>>> different systems interfaces, and I believe a design was going to be 
>>> written on this topic by Deepak. The name should probably come from 
>>> the Systems interfaces that are provided. I'm not sure why we would 
>>> need to hardcode these. How these interfaces get created, and what 
>>> they map to, needs to be discussed.
>>>
>>>
>>>>
>>>> here is the patch for code review:
>>>> https://gerrit.openbmc-project.xyz/c/openbmc/bmcweb/+/32971 ;
>>>>
>>>> for example, Node1:
>>>>
>>>> /​redfish/​v1/​Systems/​{ComputerSystemId}/​LogServices/​ 
>>>> {LogServiceId}/ ​Entries/​{LogEntryId} -->
>>>> /​redfish/​v1/​Systems/​Node1/​LogServices/​{LogServiceId}/​Entries/​ 
>>>> {LogEntryId}
>>>>
>>>> Best Regards
>>>>
>>>> Daniel(Yugang)
>>>>
>


More information about the openbmc mailing list