Next Level Reliability?

Ok so I’ve got a HE (c5) running about 100 Sengled Zigbee bulbs, 7 buttons, 9 motion sensors, a handful of door sensors, and 2 z-wave devices. I love everything about HE but as I’ve grown, I’m starting to see some reliability issues. Some things just randomly stop working then start working again. I’ve reviewed the “how to build a solid Zigbee network” article and have built things as reliably as I can. I had 5 outlet repeaters before and after it was suggested I add more, I added another 6. Things have gotten better but it got me thinking, what’s the next step for me as I continue to grow my smart-home in number of devices and complexity of automations, etc?

Would adding more hubs help? I was thinking about getting two more and having one hub on each level of the house? Is there any benefit to this? Any other suggestions for me? I’m wondering if the one hub handling everything is eating up the hubs resources?

I’d love to feel like I’m growing into something instead of out of it.

Thanks for any advice you can give!

The benefits of adding hubs is not a simple answer, although there are "bumper stickers" that might apply, such as; "It can't hurt!" :smiley:

I have 7 hubs so, understand that I'm in favor of multiple hubs. I've had three hubs interconnected for almost 2 years, first with HubLink/Link to Hub for a couple months and then HubConnect. I've started to add a 4th to the production mix, a C-7 because my original set of a C-3, and a pair of C-4's might see a benefit. I've gotten quite slow at migrating recently as the first 6 devices I added to the C-7 were indistinguishable from running on the C-3. Curbed my enthusiasm. :smiley:

Three hubs have been quite good for me, one as "Upstairs" (ZWave & Zigbee) one as "Downstairs" (ZWave only). The Third has both radios disabled and acts as a 'coordinator and internet facing hub'. I have Dashboards and weather and Voice (Alexa & Google) as well as a connection via Homebridge to Apple (and Siri.)

So, my kneejerk reaction to the question is: Yes, having the devices split means 'more headroom' in the resources of each hub. It also means you lose only half the house, should one fail. If nothing else, it's handy to have a spare without having to wait for delivery.

1 Like

If you go with adding a second hub, try to localize the devices around each hub so there are less/no hops back to the hub.

You should have 1 repeater for every 10 to 20 or so Zigbee devices. The repeaters vary in quality a bit so it really depends. I have 1 for every 6 and have never had any Zigbee issues regardless of hub build.

You could also build out your zwave mesh since your Zigbee side is pretty loaded and it sounds like there is plenty of room there.

2 Likes

I see this question more and more. In reality we shouldn't need multiple hubs, but a number of people are facing reliability issues, slow downs etc.. I guess a new hub can't hurt, although I am not in favour of multiple hubs myself. I'd like one that worked.

Fortunately until recently I have been in the happy place, but my hub Seems to be slowing even though I have offloaded rules to Node Red, but it's not awful yet.

1 Like

Multiple hubs are actually a great idea in certain situations. If you have lots of lightbulbs, a second hub dedicated to those bulbs ONLY is a great way to speed things up. Bulbs are terrible zigbee repeaters, and keeping them on their own network is crucial if you have enough to be worth a second hub. I moved my bulbs (only ~25ish) to a second hub with great success. I would also say that the hub rebooter app is great. I have mine rebooting every week on a schedule and I have 0 slowdowns anymore.

Second hub is pricey over hue though.

1 Like

For my client I added a second hub (a C-5) to their detached garage which is a fair distance from the house. Fortunately they have a cat 6 run to the garage so all good. I think multiple hubs really do help with longer distance devices especially with Z-Wave (and the current growing pains). The house itself only has a single C-7 but lots of repeaters. We'll see how that works out - Built in 1800s / 3 floors + basement / +3500 sqft.

For my house I originally had 3 hubs - Upstairs C-4, main floor C-4, and a C-5 "control" hub (internet stuff/misc apps). I am migrating my Z-Wave devices to a C-7 and possibly a C-5 for Zigbee. I really liked having separate hubs though. As long as I kept the rules localized to each hub an issue with one did not impact the other.

1 Like

I've never liked the idea of multiple hubs of the same kind, ie. Two Hubitat hubs. I currently have one C7 and one Hue bridge. I believe one hub should serve the purpose, with exceptions of course.

My house is 2700 ft2, built in 2014, two story brick. Nothing special except that all interior walls are insulated.

The Hue bridge has 47 bulbs connected. Hubitat has 43 swabs and 53 Zigbee devices. There are around 50 virtual devices, I use EchoSpeaks alot and all of my automations are handled via WebCore. I also have devices outside.

The range on the C7 is much better than any of the previous hubs. I believe this alone has greatly improved the hub reliability.

My automations have been very reliable. I have not experienced any problems at all really, and by that I mean the hub seems to have issues when upgraded. It seems to have mesh issues afterwards, but a good solid hard reboot takes care of most issues.

Overall, I'm very happy with the C7.

I'm posting this to give another perspective.

1 Like

I have 1 C7 fairly evenly loaded with zigbee and zwave. 220ish devices in a 3400sq/ft 1 story home. All my frequently executed rules I've kept as simple as possible. I do have a couple of big rules for going to bed and waking up but the rest of the time it's all pretty straightforward. The hub is in the middle of my home and has a pretty open floor plan. There's no perceivable lag with any of my automations. I opted for Google Home for voice control over Alexa because there was about a 200-millisecond difference between the two, with Google Home being faster. I also prefer the Community Google Home Integration app over the built-in.

Virtual avg: .041
Z-Wave avg: .51
Zigbee avg: .116

2 Likes

I dunno the redundancy of having 2 or more hubs is a nice safety feature and it definitely reduces overhead. Also nice to be able to reboot/update one hub and not cause issues with the others. The trick is in managing the design of such a multi-hub scheme though - how your rules will work, how to connect the devices across hubs etc.

Also older houses can have wonderful things in the walls like metal lath that can interfere with signal propagation or in my client's case the garage was just too far away to be reliable and yet they wanted automation of lights and various things in the garage.

In addition I am a big fan of isolating cloud applications and such or reducing their use if possible. Just not sure you need a separate hub for just this - I had one but am now getting rid of it as I have moved the cloud stuff to other servers.

I agree with you though (and as others have said many times) that one hub should be enough for most people and that is why I am making the effort to eliminate most of my hubs..

We have an 8 core CPU in the hub... don't hear a lot of suggestions that it be reduced to a single core. Or to go with a double speed CPU with only 4 cores. Each Z-Device has a tiny CPU.

In other words, parallel processing is a given in just about every Home. Invisibly perhaps, but there's good logic in segregating processes, in my opinion. :smiley:

The same "one hub should be enough for most people" logic applies to Phones too, right? One Phone per house, a la 1960's, with the long curly cord? :smiley: Remember when we all stepped up to two lines in the house? One for the modem :smiley:

1 Like

So are you advocating for me to use a single hub or add one or more? :slight_smile:

Also, I am on a C5 revision, do they all have the same specs as far as processor, cores, memory, etc, or is there benefit for me to upgrade from the C5 to the current C7?

Thanks!

:grin: I hear you!

You know I don't really know the limits of the hub. When I jumped to using multiple C-4's and a C-5 it was due to the usual firmware issues and I never really looked back.

Now that I am actually recommending it for clients and trying to keep costs low for them I want to get a better handle on the hub's capabilities so am migrating all my Z-wave stuff (upstairs/downstairs/basement) over to a C-7.

I also have a ton of zigbee devices on my main C-4 so after eliminating all Z-wave devices am going to do the backup/restore trick on a C-5. I am doing this because I am too lazy to (re)pair them to the new hub and my understanding is the C-5 has the same Zigbee radio anyway.

My house and my client's house is working really well aside from some teething pains with the device pairings etc.

1 Like

I think, today, not so much of a benefit... ZWave is an evolving spec and the C-7 hits the most modern of them.. but there will be future changes. The day to require a C-7 is in the distant future when you can't buy new ZWave devices that are NOT S2. That's the point where the older non-C7 will begin to age out. Somewhere between today and that day is when you (and I) should upgrade.

I have always used multiple hubs, since before I purchased my first Hubitat, and I can't imagine the day I won't. But that's not the same as saying it would benefit anyone else. I think there's a large subset that can benefit, but don't like the added task of remembering where a device is located.

I had that experience yesterday... someone around here asked about RM's Global Variable and mirroring it via HubConnect. I know I tested that, and I know that I've got a test device/connector.. but I hadn't used it in at least 6 months.. I couldn't remember which hub had the 'original'. :smiley: It's an unusual case for me because I've got my house physically split to each hub. Thus if the device I'm looking for is "Upstairs" there's only one hub it could be. But for virtual devices.. could be anywhere and often is.

I'm a believer in multiple hubs... there are some big glaring benefits and smaller.. some are barely detectable. One I use often is "Compare" -- where there's a question about an automation and it's just so easy to create it again on another hub to compare it for differences.

1 Like

Without data it's really hard to say weather or not you need a second hub.

http://hubitat.local/hub/enableStats (Run this to start stat collection and let it collect for about 5 minutes)
http://hubitat.local/hub/stats (Run this after the 5 minutes has elapsed to view the stats)
http://hubitat.local/hub/disableStats (Run this to stop stat collection)
For Winodws PC's Shift + Windows Key + S is an easy way to take a cropped screenshot.

You can use these links to track down the apps and devices by putting the id's from stats
http://hubitat.local/installedapp/configure/APPIDHERE
http://hubitat.local/device/edit/DEVICEIDHERE

The info from this can help guide your decision.

2 Likes

Cool! I had no idea about this! I ran it quick but it's worth noting, I am home alone working right now. The other 5 people will be home tonight so I will run it again while everybody is here. That being said, here is the data. Anything standing out?

Device Stats enabled: true
Device stats start time: 1605557400593
Device stats total run time: 1515119
device id 238 runcount 27 total runtime 526 average run time 19.4814814815
device id 644 runcount 23 total runtime 272 average run time 11.8260869565
device id 63 runcount 48 total runtime 1398 average run time 29.125
device id 33 runcount 5 total runtime 30649 average run time 6129.8
device id 623 runcount 13 total runtime 249 average run time 19.1538461539
device id 645 runcount 9 total runtime 172 average run time 19.1111111111
device id 68 runcount 7 total runtime 85 average run time 12.1428571429
device id 449 runcount 6 total runtime 59 average run time 9.8333333333
device id 740 runcount 1 total runtime 33 average run time 33
device id 642 runcount 4 total runtime 26 average run time 6.5
device id 616 runcount 5 total runtime 43 average run time 8.6
device id 588 runcount 4 total runtime 164 average run time 41
device id 585 runcount 10 total runtime 106 average run time 10.6
device id 586 runcount 10 total runtime 83 average run time 8.3
device id 587 runcount 10 total runtime 114 average run time 11.4
device id 99 runcount 20 total runtime 182 average run time 9.1
device id 366 runcount 2 total runtime 86 average run time 43
device id 365 runcount 6 total runtime 71 average run time 11.8333333333
device id 364 runcount 6 total runtime 50 average run time 8.3333333333
device id 363 runcount 6 total runtime 60 average run time 10
device id 390 runcount 10 total runtime 112 average run time 11.2
device id 578 runcount 4 total runtime 223 average run time 55.75
device id 547 runcount 10 total runtime 110 average run time 11
device id 546 runcount 10 total runtime 134 average run time 13.4
device id 577 runcount 10 total runtime 97 average run time 9.7
device id 548 runcount 10 total runtime 112 average run time 11.2
device id 611 runcount 10 total runtime 138 average run time 13.8
device id 393 runcount 3 total runtime 109 average run time 36.3333333333
device id 389 runcount 7 total runtime 72 average run time 10.2857142857
device id 391 runcount 7 total runtime 74 average run time 10.5714285714
device id 392 runcount 7 total runtime 63 average run time 9
device id 233 runcount 1 total runtime 17 average run time 17
device id 432 runcount 1 total runtime 41 average run time 41
device id 417 runcount 2 total runtime 18 average run time 9
device id 425 runcount 2 total runtime 17 average run time 8.5
device id 418 runcount 2 total runtime 17 average run time 8.5
device id 426 runcount 2 total runtime 17 average run time 8.5
device id 142 runcount 4 total runtime 34 average run time 8.5
device id 434 runcount 3 total runtime 108 average run time 36
device id 419 runcount 8 total runtime 60 average run time 7.5
device id 420 runcount 9 total runtime 75 average run time 8.3333333333
device id 124 runcount 12 total runtime 88 average run time 7.3333333333
device id 383 runcount 1 total runtime 45 average run time 45
device id 384 runcount 1 total runtime 82 average run time 82
device id 380 runcount 3 total runtime 24 average run time 8
device id 388 runcount 3 total runtime 28 average run time 9.3333333333
device id 381 runcount 3 total runtime 25 average run time 8.3333333333
device id 513 runcount 3 total runtime 23 average run time 7.6666666667
device id 234 runcount 1 total runtime 12 average run time 12
App Stats enabled: true
App stats start time: 1605557400596
App stats total run time: 1515129
app id 586 runcount 28 total runtime 744 average run time 26.5714285714
app id 610 runcount 14 total runtime 474 average run time 33.8571428571
app id 123 runcount 42 total runtime 2626 average run time 62.5238095238
app id 111 runcount 10 total runtime 1026 average run time 102.6
app id 547 runcount 8 total runtime 215 average run time 26.875
app id 590 runcount 10 total runtime 883 average run time 88.3
app id 615 runcount 9 total runtime 694 average run time 77.1111111111
app id 357 runcount 4 total runtime 109 average run time 27.25
app id 114 runcount 24 total runtime 1788 average run time 74.5
app id 545 runcount 8 total runtime 312 average run time 39
app id 224 runcount 2 total runtime 209 average run time 104.5
app id 374 runcount 6 total runtime 153 average run time 25.5
app id 387 runcount 2 total runtime 49 average run time 24.5
app id 396 runcount 1 total runtime 124 average run time 124
app id 389 runcount 6 total runtime 151 average run time 25.1666666667
app id 513 runcount 8 total runtime 600 average run time 75
app id 369 runcount 2 total runtime 57 average run time 28.5
app id 370 runcount 1 total runtime 68 average run time 68
app id 115 runcount 13 total runtime 1056 average run time 81.2307692308

Device ID 33 is suspect due to the small number of runs yet the large total runtime. Devices are usually around 10 or 20 milliseconds.

Additionally, you might want to look into slowing down the chattiness of your high run count devices and apps.

Cool. I killed that one. It was my Roku TV. Also another LG WebOS TV looked suspect too so I killed that too. Like you, I like making decisions based on stats. I'm glad I know how to run those now. Are there any other resources regarding running/reviewing stats on the Hubitat to continue to tweak things?

Yes, there are a lot of hidden features but hub memory and stats are the most useful to me.

1 Like

I have a few suspect apps but I know what they are and what they are doing, I've taken steps to slow them down, and can disable them if they ever cause me an issue. They're the first ones I disable when I am performing any troubleshooting just to quickly rule them out.

1 Like