Spanning tree states

  • 10 February 2009
  • 13 replies
  • 2840 views

How often does the system recalculate its spanning tree states? And does it take into account link quality (error rates) in its route cost calculations?

I have a troublesome zone ('Kitchen'), which is sometimes prone to dropouts. When I looked at the Network Matrix I find that the ZP in question is currently forwarding to a ZB which is relatively remote and at low strength ("Inbound: 24 21; Outbound: 04 19"). It has nodes in range with much better signal strength, and indeed one of these ("Inbound: 54 56; Outbound: 42 46") is currently forwarding to it. But 'Kitchen' itself insists on forwarding on a weak signal link.

I've tried changing channels, power cycling various nodes in different orders. Will it just adjust itself after some period of time? I'm hoping someone knowledgeable in SonosNet can help me out. Forgive me if this is on the wrong forum.

13 replies

Hmmm... Today 'Kitchen' is forwarding on a much more sensible link ("Inbound: 43 47; Outbound: 51 56") to a nearer ZB. So I guess that partially addresses my query. I'd still like to know how SonosNet chooses between links, and how often....
Userlevel 2
i have often wondered the same thing. for some reason my kitchen seems prone to "making the wrong choices in life" too. must be something about being exposed to cooking odors. don't know if it would help anything in reality, but might be nice to have a "nudge sonos to look for best connection" option in system config.

J
Not so much cooking odours, as cooks getting in the way of the signal? Today the kitchen ZP was happily forwarding on a link with signal level "Inbound: 40 38, Outbound: 51 52". Then when I stood near it the music dropped out as the link reconfigured itself to a different ZB at "Inbound: 30 30; Outbound: 25 28". I can only assume I somehow shadowed the ZP from the stronger signal.

Anyway in order to provide some redundancy I've deployed a new ZB nearby today (the one to which the kitchen ZP re-homed itself as noted above) so I'll see how things go with various channel selections.
Badge +2
I have noticed that whenever the Microwave is active in the Kitchecn the ZonePlayer will start dropping the connection. I currently have the ZonePlayer about 18" above the microwave.

Could this also be causing the problems that you describe?
Could this also be causing the problems that you describe?
No, I'm talking about the criteria and frequency of optimum link selection by Sonos under "normal" conditions.

The microwave is another - albeit predictable - problem. That completely hoses the music stream irrespective of which box the kitchen ZP is talking to, and often Sonos won't recover and I have to restart play from the controller. (Interestingly a Squeezebox I used to have didn't suffer from this problem despite being in the same location.)
Userlevel 2
I'm not sure if in this case you can force a route (and get it to always route as you've told it). One possible solution would be to wire the zone player. This could be a pain since its in you kitchen but the devices that allow you to pass the signal thought the electrical outlets have worked for me in the past. If you can't get it to work the way you want you might want to consider this option.
I'm not sure if in this case you can force a route (and get it to always route as you've told it). One possible solution would be to wire the zone player. This could be a pain since its in you kitchen but the devices that allow you to pass the signal thought the electrical outlets have worked for me in the past. If you can't get it to work the way you want you might want to consider this option.
Thanks for this suggestion. I've considered the idea of using powerline adapters to connect to the kitchen as Plan B. Do you happen to know whether (a) a ZP will favour a wired connection via the electrical outlets over a wireless link when the latter is normally quite strong (other than when the microwave is on of course), and (b) whether latency via the powerline adapters is noticeable when zones are linked together?

Also any recommendations for powerline adapters would be most handy. Thanks.
[ ... ]
a ZP will favour a wired connection via the electrical outlets over a wireless link when the latter is normally quite strong (other than when the microwave is on of course), and (b) whether latency via the powerline adapters is noticeable when zones are linked together?


SONOS will use the lowest latency connection. This will usually be the wired connection, but the power line adapters will tinker with the timing.

The SONOS networking will automatically deal with the latency, linked zones will always be in sync.

If one measured the absolute delay from a Line-In to a ZonePlayer output, the delay may vary slightly from network to network, but this variation would be very small compared to the Line-In encoding delay that is intrinsic to the process.
My kitchen zone drops off if I open the fridge door.

I've lost 20lbs since installing it.

T
SONOS will use the lowest latency connection. This will usually be the wired connection, but the power line adapters will tinker with the timing.
Thanks buzz. In the event that the power line route had additional latency (and therefore was less favoured) do you happen to know whether the ZP could retain that route as a backup, and switch over seamlessly when its RF transceiver is temporarily knocked out by the microwave?

BTW, a bigger problem than the temporary dropout - nuisance though that is - is the fact that when the link recovers the ZP starts declaring loads of tracks corrupt and often skips back to the start of the playlist.

The SONOS networking will automatically deal with the latency, linked zones will always be in sync.
Yes, thinking about it I guess they use some form of timestamping of the music frames and periodic clock sync between players.
[ ... ] do you happen to know whether the ZP could retain that route as a backup, and switch over seamlessly when its RF transceiver is temporarily knocked out by the microwave?


I don't know that detail, but why bother? They shouldn't assume that the prior route would automatically be better than the current route. (If it was, why are we here?) If a link goes down, they must build another path. The process can be accelerated if they periodically probed and kept track of "neighbors". Continually plotting all possible routes is process intensive, tracking adjacent neighbors is relatively cheap.

[ ... ] the ZP starts declaring loads of tracks corrupt and often skips back to the start of the playlist.


This suggests that the path to the music was broken. SONOS skips the "missing" track and moves on to the next. When this happens, you could check the IP addresses of the ZonePlayers. Probably the player associated with the controller is isolated.

The diagnostic report has all of the information that you need, somewhere, but it is not documented.

/opt/log/anacapa.trace
/proc/ath_rincon/phyerr

will be interesting for you.
I don't know that detail, but why bother? They shouldn't assume that the prior route would automatically be better than the current route. (If it was, why are we here?) If a link goes down, they must build another path.
Okay point taken. I'll give it a try with a pair of powerline adapters to that ZP.

The diagnostic report has all of the information that you need, somewhere, but it is not documented.

/opt/log/anacapa.trace
/proc/ath_rincon/phyerr

will be interesting for you.

Yes, I've found /proc/ath_rincon/phyerr is useful, and correlates with the timing of the interference bursts.

/opt/log/anacapa.trace appears to be less useful - just a load of sync errors reported. The timestamps are completely wrong too, though one can infer rough clock time from the track titles in the scrobbling submissions.
I'll give it a try with a pair of powerline adapters to that ZP.
Just to follow up. I installed a pair of devolo HS dLAN adapters to provide a wired link to the troublesome ZP and - fingers crossed - it's been behaving impeccably. The devolos report 60-80Mbps between the adapter in the kitchen and the one connected to a switch upstairs. They have no problem with being plugged into standard 4-way extension blocks.

Reply