[mh] Does hops count in request status count going through APs?

[Marc M. <ma...@me...>]

I've mostly given up trying to debug my setup. It seems to work well when I
don't do very heavy activity like a full scan or sync of everything.

While debugging though, I found another test: 
I would initiate a request status to my fmr_kpl and look at hop count.
Typically it would be 2 all the time, and sometimes it would dip to 1 or 0
for several times and then come back to 2 for a while.

This could coincide with a fridge or thermostat, but I was able to reproduce
the occasional 1s and 0s even with 80% of my house disconnected from the
circuit breaker panel, including most plugs and all fridges, 220V devices,
photovoltaic system, and so forth.
It's very vexing because after hours of trying I cannot see what would stop
my signal from getting to destination on the first send, or first
rebroadcast, and only sometimes: if my signal was too attenuated it would
fail a lot more often, and if it was a rogue fridge or power supply
somewhere, those I removed from my house at the circuit breaker. 
I'm starting to think that it's because of random AP bridging problems due
to RF, except my house seems to bridge without any APs, and I have 4 of them
with 2 on each phase...

That said, the good news is that even with hop count 0, I never got my
request dropped, so that's why my house has otherwise been working ok.

Now, 3 questions:
1) hop count gets decreased on each rebroadcast. My understanding is that 
each command gets sent everywhere, then all the devices rebroadcast the same
signal something like 0.3s later, and then then rebroadcast it a second time
0.6s later or so.

The PLM then states which response it got: first one or 1st or 2nd
rebroadcast. Getting hop count 0 actually reflects that occasionally the
path from my KPL back to my PLM can be a bit too long or attenuated, and
require one or 2 rebroadcasts to get there (yet, I only have a 2600sqft
house, and my failures are not on the farthest switch anyway).

Am I correct that going through an AP is done real time and has nothing to
do with rebroadcasts or hop counts?
(but if so, how can a receiving AP, especially in my house where I have 4
APs, rebroadcast on the power line at the exact same time than the signal
that's already travelling without creating a new powerline signal that's
offset somewhat? My guess of an answer is that the powerline signal is 
low bitrate enough that retransmitting it 0.001s later doesn't really change
it enough to matter)


2) In my house, it looks like occasional hop count 0 on my KPLs will simply
delay the signal somewhat (0.8s -ish?), but as long as the 2nd rebroadcast
reliably reaches the other end, I should be ok.
Right?

Then, if that doesn't work, my KPL should resend the original signal twice,
but with a bigger delay this time. Correct?


3) If I have a KPL talk to another one, KPL 1 has to wait for 1 second to
get its ack (time for 1st signal plus 2 rebroadcasts).
Then, if it doesn't get an ack, do KPLs and switchlinks resend the signal
twice like mh does, or if the first transmit doesn't get there, the command
is lost? (I think it's the former)

If it does try to resend twice like I think it does, then if I get this
properly the signal has 9 chances to get there (3 sends with 2 rebroadcasts
each) and in my case, probably 18 chances since I have 2 APs on each of my 2
phases that do their own synchronous rebroadcasts too.

In other words, even if I get hop count 1 or 0 or maybe even the occasional
retransmit, I should still have many changes for the signal to get there,
right?


I'll update the wiki with the information I get since understanding what's
behind the technology can be essential to proper debugging.

Marc
-- 
"A mouse is a device used to point at the xterm you want to type in" - A.S.R.
Microsoft is to operating systems & security ....
                                      .... what McDonalds is to gourmet cooking
Home page: http://marc.merlins.org/