🚀 go-pugleaf

In article <40056@fibercom.COM> mack@frugal.fibercom.com (Richard Mack) writes:

>Could you please post the situation where a port could pass LCT and not pass
>packets? The only thing I can envision would be if the PH_DATA.(request||data)
>lines in the concentrator are broken, but the station self-test should be
>able to find that.

Here are a two others:

1. LCT is a PHY level test. It does not address the MAC's ability
to move frames. It is conceivable that, in between the time since
the DAS node performs its self test before applying for entry
to the ring at either concentrator and the time when a link
to one concentrator fails and it has to connect through the other,
the DAS no longer has a good connection between the MAC and the PHY.

2. LCT exercises the link with line states. Since these are not long
frames, they do not stress the elasticity buffer. Thus, if a clock
on the DAS or the concentrator were out of spec, you would not
know it until you were actually using the link to transmit
long frames.

PLEASE do not misread my citing of these pathological cases as
any lack of support for dual homing. I believe it is a valuable
mechanism for enhancing system availability. This is just a
response to Richard's request for hypothetical situations.

By the way, I believe a DAS will preferentially connect through
the B port when given the choice. The Committee decided to do
it this way so that management nodes can draw deterministic
ring maps. I mention this because of the clocking issue
I cited in case 2. If a concentrator connects only to the A ports
of DAS nodes whose B ports are all satisfied by a second concentrator,
and the first concentrator's clock is out of tolerance toward the
fast side, and the two neighbor nodes to the first concentrator were
within tolerance toward the fast side, then you would discover the
problem in a nasty way if the second concentrator failed and
forced all the DAS nodes to connect to the first. Now you will
get occasional elasticity buffer overflows on long frames.
Perhaps a better way to manage a facility is to set up dual homing
connecting half of the concentrator's M ports for A ports and the
other half to B ports... of different stations of course ;-).

Please also note that oscillators drift with time and temperature
so even if you are able to pass long frames within a minute or two
of powering up a device, that does not mean you will continue to be
able to do so after the equipment has been cooking away for a few
hours on a hot day.

Ron

Ronald S. Perloff     rasp@xlnt.com
XLNT Designs, Inc.  (XDI)
15050 Avenue of Science
Suite 106
San Diego, CA  92128
619 487 9320   -   FAX: 619 487 9768