Frequent pauses leading to a silent device. Power cycling is the only workaround.

Do the garden offices have an Ethernet connection? Where are the Play:1s with disabled radios located?

It would also be useful to see a screenshot of the Network Matrix, to be found at http://x.x.x.x:1400/support/review, where x.x.x.x is the IP of any player. Blur out the MAC addresses if you’re concerned about privacy.

“Office” is presumably the wired Connect. “Kitchen” is obviously wired too. “Patio 2” is the only other wired node by the looks of things. 

Two nodes look to be powered off. What happens if they’re on?

When the problems start to set in, do the switch port LEDs give any indication of abnormal traffic patterns?

Do you typically have one or multiple units playing? Are any players Grouped? There may be a transient source of interference. Does the Network Matrix change significantly over time?

When paired, the left unit becomes the “coordinator” for the pair. All data for the pair is managed by the coordinator. For Groups, the first player (used to build the Group) is the Group coordinator. When play is initiated a cache is built. Data from the cache is used to ride through small communication issues. If there are lots of issues the cache will shrink and eventually the unit will need to take action when its cache is exhausted. If (in the opinion of the player) the issue seems transient, the player will briefly mute. If the problem is severe the current track will be abandoned and the next track in the Queue will be attempted. A string of small issues can result in an apparently silent player as it patiently tries to work through a continuous string of small, transient issues. Obviously, real time services, such as net radio cannot skip to the next track.

Keep a log of issues. I once had a very severe, intermittent issue. After a few days of logging it became obvious that days and occasional evenings were fine, while other evenings were nasty. Sunday afternoon was nearly impossible. I slipped a copy of my log under my neighbor’s door, along with a polite note. A couple days later the annotated log returned. Some evenings she was traveling overnight for business and Sunday afternoon she called Mom. She was using a cordless phone system that was known to wreck WiFi. I was able to completely resolve the issue by avoiding wireless SonosNet connections near her base station.

You may have an intermittent hardware issue that interferes with network communication. This may or may not be SONOS caused. In another situation network traffic was much slower than expected for a wired network. It was very puzzling. Eventually I discovered that a hidden network wire had been damaged during construction. SONOS noticed this and used its wireless SonosNet to provide the connection to a network switch. While SonosNet is solid, it is not very fast compared to a Gigabit wired connection.

Thanks. Interesting about the phantom devices. I wonder why they are there. I run my own DHCP/DNS server (DNSMasq). It has the IP address of each Sonos device along with its MAC address.

We cannot rule out the possibility of a hardware or very specific to one player network issue. For the purpose of your logging, you could systematically avoid Grouping or using each player. Submitting diagnostics could reduce your logging workload because the diagnostics log network issues for all players -- unless the situation is so bad that a player cannot report, but this is a good, though incomplete data point.

Update…..! The initial support technician told me to wire all the speakers to the router. I did this but the problem persisted.

I submitted another diagnostic. This time the (excellent!) technician noted that many of the devices are suffering from low memory. Which fits the symptoms exactly as after a power cycle they function fine until the problem reoccurs.

Most of these devices are old and have less memory than the current models. He suggested that all the Connects and the Connect Amps should be wired with the wifi disabled. This would lighten the load on the problematic devices. He also suggested that the single wired "gateway" should be a Play One, which has more memory than the Connect/Connect Amps and so would be less likely to suffer the issue.

However, the only true solution would be to replace all the devices with current models, which have more memory. Unfortunately the Port and the Amp are considerably more expensive than the Connect and Connect Amps. Perhaps I need to bit the bullet and look for Black Friday offers today!

I understand the

Is it the Play:1 that only has 64mb of Ram?

According to this chart Play:1 has at least 128MB.

A further update. After wiring the majority of the devices, including all the Connect/Connect Amps (and disabling wifi) a few days later the problem returns. So it’s looking like the Connect and Connect Amps are nearing the end of their life! That’s going to be quite an investment for me to replace them all. I’ve taken a diagnostic and will contact support with it. But it looks like an unworkaroundable issue :-(  I wonder if I can switch to S1? I use SMB stored flac files for a music library, streamed radio, Apple Music and Spotify.

I finally found time to follow this up with Sonos, with fresh diagnostic reports after their advice to wire as much as possible to reduce memory usage on the older devices.

After lots of back and forth about not wiring speakers as it causes a broadcast storm, I was able to inform the representative that only one wired speaker had wifi enabled and I had not seen a broadcast storm for years! ..AND that is was sonos support that recommended me to do this :-)

The representative then said "a system update is the best way to proceed. But we do not have an ETA".

I asked for confirmation that this was “..the memory issue on the older devices” and the rep confirmed that this was the case and repeated that  "..At this time, updating your system is the best way to alleviate this issue. We do not currently have an ETA.".

So it looks like Sonos will hopefully reduce the memory usage for these older devices with a firmware update.

Not sure what you mean by “dead memory”, but according to earlier comments from support, it’s just the 2nd generation Connect and Connect Amps that are problematic. Hopefully whatever memory leak (which it must be as this only happens over time and a power cycle fixes it!) is happening can be reduced on those devices.