Re: [Linuxptp-users] Question re: boundary clock behavior?

[Richard C. <ric...@gm...>]

On Fri, Jan 18, 2019 at 06:55:43PM +0000, Corey Stinson wrote:
> However, if port 1's grand master goes away due to, specifically,
> port 1 being disconnected from its switch, wire physically cut,
> etc. this (correctly) reverts port 1 to Fault state.  According to
> IEEE 1588 specifications, such a fault shall not translate to other
> ports on the node.  Therefore, Parent Data Sets are not updated for
> port 2 and continue to reflect the prior grand master configuration
> even though that grand master is no longer available.

Once the upstream Announce messages time out, then the Parent Data Set
will revert back to the BC's default data set.

> I feel as though I must be missing something obvious?

You forgot the announce timer on the BC.


Thanks,

Richard