I have the Hub Watchdog app installed with two child instances, one monitoring a Z-wave smart plug-in switch and another monitoring a Zigbee smart plug-in switch. App is set to run every 3 hours and log the response times for Z-wave and Zigbee. This basically gives me an idea of how fast my wireless mesh is responding for both wireless protocols.
What are the S2 Authenticated devices?
They are half Ring Contact Sensors 2nd Gen (doors and windows) and half GE/Enbrighten smart in-wall switches/dimmers/fan controls. All of these devices at the time of inclusion prompt to include with S2 authenticated/unauthenticated. When I first installed these devices I took the early advice I read which was "don't touch the checkboxes in the include dialog and just hit ok".
Why did I second guess my S2 Authenticated include choices?
The Hub Watchdog app has been reporting my Z-wave mesh had a median average delay of 550ms or more. There was also regular jumps to 750-800ms and sometimes weird spikes to 8-10 seconds. I have noticed at times some noticeable delay in automations firing and controlling Z-wave in-wall light switches which I contributed to these Z-wave network delays.
What did I change?
In batches of about 6 per day I went around and excluded each of the S2 Authenticated devices and included again, this time unchecking all the dialog boxes at prompt. This transitioned devices from "S2 Authenticated" to "None". I also worked my way from the hub out into the rest of the house.
Final results
Once finished with transitioning all devices to "None" the Hub Watchdog app is now telling me my median delay is down to 380ms and very rarely I see spikes above 500ms. No more weird 8-10 second delays appearing so far, knock on wood. More importantly I can tell a noticeable difference in the responsiveness of automations controlling the Z-wave devices.
You may ask, what about that single S0 device? That is a older Z-wave only Schlage door lock that is getting replaced this month. It will join the other three Schlage door locks as S2 Access Control here soon.
I hear you, but I was bored and wanted to see if the effort would make a improvement in my network, which it did.
Also since initially buying the HE and installing these devices I have done a lot more reading here in the Community. What I learned was S2 authenticated, while default for some devices, is really not needed and results in unneeded packets across the mesh network. The only place where authentication is really recommended is smart door locks (S2 Access Control) which I will leave in place.
Security is never "needed". It is always subjective depending on use and function. Just like passwords, accounts, locks, etc.
And "recommended" is subjective, too.
Not really / only in rare circumstances where the packet size limitation comes into play. With S2 anyway, which was the context. S0 is another story.
Regardless, interesting findings. Thanks for sharing (sincerely). Hopefully those root issues that caused your slowdowns will get ironed out (for others).
That was the goal, simply share some data. I knew some of this would be contentious in the group, but if anything new users discover there are tools to monitor the health of your Z-wave wireless mesh network. Then choices you make can change the results.
3-week update. I have changed every device on my home Z-Wave mesh network from S2 to None. I've also removed the single S0 device by replacing it with a device that supports S2 Access Control like the other three.
The results, my mesh has been super solid, super responsive, and overall well worth the effort and changes. To that point here is a snapshot this morning of my Hub Watchdog for my Z-Wave mesh response times. Before I would get an average around 800ms and wild spikes throughout the day that could go as high as 10 seconds. All of that is gone, and it's settled down to an median of 475ms.
I've also started a thread for others to share their Hub Watchdog results so we can all get a sense for what is fast and what is not, where is the benchmark?
I have a 2 story w/ full basement about 3600 sq/ft
I located the hub in the most central location of the main floor I could and worked from the hub out. I initially started with the plugs since it was quicker to put them in place, I joined them to the hub then proceeded to install the switches (and those 2 battery remotes last) ... I set everything up using smart start with QR code so S2 security authenticated thinking this was the ideal configuration.
A few weeks passed and I was not happy with the response times. I was hoping maybe things would settle in or running repair might improve but it never did - I didn't install any watchdog service or anything to measure, just obvious. Mostly noticeable using Alexa and some physical switches with virtual 3-way and 4-way when not connected directly to the load (FWIW - I've used the mirror app and direct association for the virtual ~ doesn't seem any different either way so sticking with mirror app for now).
Over the weekend I reset everything (full reset on the hub, exclude w factory reset on the devices) and rejoined (with switches first this time since they were already wired, dont know if it that matters) then plugs, then the battery remotes. This time I did everything with no security. I can say my experience is very much like yours - response is markedly improved. Even with Alexa it is now turning on/off devices as soon as I utter the last syllable!
edit: forgot to mention I had already updated the hub to latest release / z-wave firmware before reset, also zigbee disabled before and after since I'm strictly using z-wave
edit 2: mirror app was getting flaky on me - went back to direct associations
yep finding this out now - kinda wish I knew out of the gate but all's well that ends well
no ghosts before - actually do have 1 now from a plug in my garage (not a door opener, its for a heater) that I still need to get up on a ladder to reset, but its obviously not causing me any issues
I definitely will - I could force remove it now but wanted to do the exclude, reset, rejoin routine. By not causing issues I wasn't assuming it never would I only meant perceived performance at the moment
If it's dropped off the mesh, excluding it will not get rid of the ghost. It's not like zigbee. It's possible if you power cycle the plug it will find the mesh again but if it doesn't you will have to rip the ghost out by it's roots so to speak
It is always subjective depending on use and function. Just like passwords, accounts, locks, etc.
