Chassis reset

[Ed Tanous]

On Wed, Sep 23, 2020 at 3:01 PM Bills, Jason M
<jason.m.bills at linux.intel.com> wrote:
>
>
>
> On 9/23/2020 2:42 PM, Patrick Williams wrote:
> >
> > I'm not understanding what you mean by "come up with an API to steer the
> > Redfish..."  I think everything is specified here at a dbus level.  The
> > issue is figuring out the appropriate Redfish model of
> > Chassis/ComputerSystem objects (along with the included Resource.Reset
> > types).  To a casual reader, who hasn't been involved much in Redfish
> > implementation, the current mapping of these ResetTypes seems fairly
> > arbitrary.
> >
> > With CIM there use to be these Profile documents that showed "the right
> > way" to fit all these pieces together.  Does that not exist with
> > Redfish?  How does any external application consume Redfish in a
> > consistent way?
>
> I'm not sure it helps with the overall consistency question, but for
> OpenBMC, we defined our mapping of Redfish ResetType values to Chassis
> and Host State Transitions on D-Bus, here:
> https://github.com/openbmc/docs/blob/838fa962b73deb1ef7908ee8dfa1203fa5263517/designs/state-management-and-external-interfaces.md#proposed-design.
>
> This is what x86-power-control aspires to match and I think what led to
> creating new D-Bus interfaces for the chassis_system and AC power state
> changes (since we didn't want to break this mapping).

Sounds reasonable, it just makes things slightly more complex for
Redfish, as now there's two things that could be a redfish Chassis
reset.  Not a huge deal.

>
> Perhaps we need to separate the Host and Chassis on the Redfish side and
> extend the D-Bus interfaces to support all the needed ResetTypes?
> >

I'm not following.  There's a Host (ie ComputerSystem) and Chassis in
redfish today.  How would we separate them further?