Re: [Linuxptp-devel] [PATCH 1/4] Schedule announce TX timeout also when in slave-only mode

[Keller, J. E <jac...@in...>]

On Sun, 2013-12-01 at 10:57 +0100, Delio Brignoli wrote:
> On Dec 1, 2013, at 7:48 AM, Richard Cochran <ric...@gm...> wrote:
> 
> > On Sat, Nov 30, 2013 at 09:46:46PM +0100, Delio Brignoli wrote:
> >> 
> >> Yes, but... 802.1AS does not have a different state machine for
> >> slave-only operation, instead a non grand-master capable time-aware
> >> system has a property1 value of 255 (see second paragraph of section
> >> 8.6.2.1). Additionally, according to point f.5 of section
> >> 10.3.12.1.4 updtRolesTree(), if the current portPriorityVector is
> >> the portPriorityVector of the time-aware system (i.e. we have not
> >> yet received an announce message with better portPriorityVector) the
> >> portRole should be set to MasterPort which makes the txAnnounce
> >> method reachable.
> > 
> > This brings to mind two questions.
> > 
> > 1. Does omitting the announce message from a slave only node cause you
> > interoperability problems?
> 
> Yes, we are failing interoperability tests without this fix.
> 
> > 2. What is the point of the slave only node's announce messages?
> > 
> > [ The announce messages will not affect the outcome of the peer's
> > BMC election, except to make it slower. ]
> 
> 
> Knowing the node exists and it is a slave I guess...
> 

I believe it's in order to know that the node is asCapable.. I know at
least one switch which will ignore a port unless it receives an announce
from it that indicates it's asCapable...

Regards,
Jake

> >From point i.4 of section 7.5 "Differences between gPTP and PTP": "all time-aware systems are required to participate in best master selection (even if it is not grandmaster capable)".
> 
> Thanks
> --
> Delio
>