WAX APs intermittently reporting "resumed connection with the NETGEAR Insight etc . . . ."

I have 6 APs - 5x WAX615 and 1x WAX620.

All units are wired to a Ruckus network switch. No wireless mesh.

Internet connection is AT&T fiber via a Sonicwall TZ470 firewall.

All devices managed via Insight.

The system ran flawlessly for about 5 months until the beginning of October 2022 when I started receiving notification emails from the Insight service saying that AP so-and-so had "resumed connection with the Netgear Insight service".

No notification that connection had ever been lost and apparently no impact on local network functionality.

Sometimes several notifications per day from different APs. No identifiable pattern.

All APs are running current firmware and there have been no changes to the network.

I have been communicating with tech support, but not making much headway.

The initial recommendation was to switch from AT&T's DNS servers to Google at 8.8.8.8.

This seemed to help a little - on further investigation I confirmed that AT&T's DNS servers are really poor - sometimes unresponsive for hundreds of ms.

But the problem remained.
Looked at logs and saw a lot of scanning going on in the APs - switched all APs to fixed channels.

That seemed to help a little more - now down to one notification every couple of days.

But they are not gone.

Tried downgrading to previous code with no effect.

Now trying to see if removing and readopting the APs into Insight helps.

My conclusion at this point is that the AP heartbeat connection to Insight is set way too sensitively with none/not enough retries and/or the APs are unable to service the heartbeat timeout due to scanning/neighborhood AP activities.

OR

The Insight service is not responding quickly enough to service the heartbeat timeout.

Either way, the fact that the log shows that the heartbeat timed out, but only generates a "resumed" notification and never a "lost" notification tells me there is something funky in the code.

This is not a mission critical problem - it's a nuisance.

I could turn off notifications, but that means that I wouldn't know when something really goes wrong and defeats one of the key benefits of a centrally managed network.

Has anybody else seen this?

Any/all wisdom would be very welcome.

Thanks