Why does OpenBMC use Avahi mDNS instead of SSDP?

[Richard Hanley]

>
>
> Was on the Redfish call earlier and this forum thread was discussed. The
> Redfish members on the call did not totally agree with this statement.
> They believe SSDP has a wider adaption than just Windows. A wider
> adaption than mDNS.  Since SSDP is already in the Redfish specification
> and has been adapted by Redfish implementations, adding a 2nd discovery
> service would damage interoperability in Redfish.
> The members on the call really wanted to encourage OpenBMC to implement
> SSDP instead.
> Posted the same to the forum thread.
>

Well that is unfortunate.

>
> >>> I want to allow the BMC admin to enable and disable the mDNS discovery
> >>> service.  How can this be done?  Options:
> >>> 1. Implement ManagerNetworkProtocol.Oem.OpenBMC.mDNS.
> >>> 2. Move OpenBMC to SSDP and away from mDNS.
> >>> 3. Re-ask Redfish to add a ManagerNetworkProtocol.mDNS protocol.
> >> I don't see a strong reason to "move away" from mDNS, but I certainly
> >> think that *any* network service should be able to be disabled by an
> >> administrator to reduce the security footprint if the service is unused.
> >>
> >> So, either #1 or #3 seems best to me.  Adding optional SSDP support
> >> would
> >> be reasonable, but likely only to be added by someone who cares about
> >> discovery in a Windows environment.
> >
> > I've appended the thread to re-ask Redfish if they would support
> > ManagerNetworkProtocol.mDNS.
> > https://redfishforum.com/thread/267/add-avahi-managernetworkprotocol
> >
> If think we are down to #1 or #2.
>

I'm trying to put myself in the mind of an administrator here, and I'd
expect most are relatively inflexible. Most hyperscalars probably have
their own discovery system, and are likely supporting that regardless of
our decision here.  Likewise smaller operations probably aren't super
interested in migrating here.

Can anyone speak to the merits of the two protocols? Is there are
technical benefit to moving to SSDP.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/openbmc/attachments/20200416/d8a11440/attachment.htm>