Hi all – just wanted to drop a note to mention it’s been a very difficult day on the east coast. Local and non-local smart apps have been running with significant lag, devices have dropped from the network becoming non-responsive, it seems I can’t even add new devices to the network… And that was before I rebooted the hub. I saw the notification that there was some high server load earlier today and that the problem was marked “resolved,” but I’m still feeling the aftershocks. Support was unaware of the problem when I reported it, and they have yet to respond with specifics… Figured I’d post this just in case other folks were having similar problems and wanted to confirm it wasn’t just them.
Just throwing in a data point, not dismissing your pain. Another east coaster here and things have been calm. I was really surprised seeing the support email. I think this was the first time I remember seeing an alert coming through before people screaming that things are failing.
Thanks for contributing the data point – no hurt feelings over here! I know it makes me feel better when the community identifies problems and reports them here. It means I can relax and let the ST team figure it out instead of pulling out all the wires and batteries.
I said it countless times that I need to keep my mouth shut so I don’t jinx my set up. I just realized that some of my zwave sensors are not reporting motion, since earlier this morning. Tomorrow will be a busy day evaluating the damages. Thankfully I have a light work day.
Is it just me? ? ? Really quiet
I feel for you!
Also on the East coast. I did not see any issues from my location. Any updates on your ST setup?
From my end, I had one z-wave sensor that stopped reporting the motion (everything else was still working) and a few zigbee. Not finding any evidence that this is really related to ST. But who knows. The fact of the matter is, that I tried to use the replace on my failed zwave sensor and instead of replacing it, it deleted the sensor completely, although I had several apps tied to it. I am pretty tired of these little “happenings” I don’t know if it’s ST, my mesh, my bad luck. But after using several hubs, I ordered one more today. Vera is one I didn’t try yet…
By this morning (Friday, July 8) all of the unusual latency I was experiencing yesterday is gone and as far as I can tell, my devices are back on the network and functioning properly.
I noticed some folks popping up on the forums lately asking about system performance and whether they should invest in the ST platform. FWIW, I roughly get about 99.99% up-time with about one device out of 50 dropping off the network every other day. The vast majority of dropped devices are GE link bulbs (was anyone surprised?) and PEQ branded open/close sensors. ST branded devices are remarkably consistent and reliable. At some point I’ll replace the cheaper hardware and switch to another manufacturer, but for now I’m okay playing whack-a-mole. Anyway, 2-cents for anyone who ever stumbles onto this thread in the future…
Just a point of clarification here…
SmartThings does use two different “shards” (transaction streams) for North America,but they are not based on where you live. Instead, they are based on when your account was added. Earlier accounts are on NA01, which is pretty much any US account added before September 2015, and a few after that.
Then that reached its capacity, and a second shard was added, NA02, late in 2015. Along with the UK shard.
You may see some system messages that have “East” in them but that has to do with where the AWS servers are located, not where you live. You could live in Los Angeles or Honolulu and still be on the east shard if your account was added in 2016.
So if you’re collecting data points, any limited issues would be based on the shard you’re on, not your location.
Just another vote for more locally run apps. When I am 2 stairs from the top before the stairwell lights come on, I might as well set up the old hub again.
I had a light turn on by itself at 12:40 am this morning. This light is not triggered on by any app or routine. I also had a double tap work by itself this morning. Had 2 Cree bulbs quit reporting status using a non-standard handler yesterday. I switched these back to the stock handler and they report fine again but i lost this ability to control dimming speed. I am running hub v1 on shard 1. Nothing major but several little things. Lag didn’t seem to be a problem.
Mistrust in system
Wow, and I thought I screwed up my hub somehow. Double tap isntalled itself, my zigbee presence sensors arrived around 12:30 last night. My device count jumpped from 124 to 134 inexplicably. I am unable to include a failed motion sensor, yet a bunch of apps that were tied to it are now in unhealthy state. I haven’t had such a bad day in months! I try to fix one thing and find another that needs my attention. I think I am just going to leave it alone for now and wait for the chit storm to pass…
Phantom Switches in the last two days
This nonsense is making me question my sanity for even trying to rebuild.
Of course these type of absolutely unacceptable issues won’t be acknowledged in the seemingly meaningless updates from @alex.
When does your Vera arrive?
I guess I’m one of the lucky few that haven’t seen anything outrageous happen over the last few days… and yes I’m out East as well, in CT.
But it could just be because I have ST relegated to relatively menial stuff. It is connected to a Schlage door lock as well as my garage doors, but those haven’t experienced any sort of uncommanded shenanigans – otherwise ST would immediately lose access to them
I have no lights that ST can control, as the bulk of my stuff is TCP, so maybe that’s related?
I have a bunch of motion sensors and cameras and other stuff but none of it has acted up recently.
As professional experimental physicist and casual programmer of 16 years, I’d like to add 2 more cents to the discussion.
Bugs are annoying, frustrating, and even potentially dangerous for those of us who connect door locks to the system. Just like anyone else, I get frustrated when performance craps out. Nevertheless, in the heat of the moment, I still accept a degree of imperfection – from my point of view, problems are acceptable if not inevitable.
When I’m working on an experiment in lab, there’s always the measurement itself and the associated measurement error. These imperfections can come from equipment or methodology, but as long as the signal-to-noise ratio is sufficiently high, the result can be good enough. For IoT platforms, the analog to a measurement is the intended function, and the analog to a measurement error is that function’s success rate. In the same way I don’t expect infinite precision+accuracy in an experiment, I don’t expect zero error rates in my ST system. No matter how well engineered the platform may be, problems are bound to crop up. The central question then is: what counts as a “good enough” result?
It’s crucial to have a high signal-to-noise ratio in an experiment, and it’s important that intended functions have low error rates. It’s been stated in these forums before, but I’ll say it again – we all have our own definition of what’s an acceptable error rate, and these varying definitions color our view of ST’s performance. Given my background as an experimentalist and that I approach IoT largely as a hobby, my threshold for “good enough” is probably on the lower side compared to most. Others here may view the technology as serving a more sensitive purpose and set their thresholds according to their specific needs (looking at you @JDRoberts ). In any case, I think it’s essential to keep this point in mind when judging the various failures that are coming up in this (or other) thread(s).
My first OS on a personal computer was Windows 3.1. When I transitioned to Win95 (and again to Win98), I remember experiencing crashes, blue screens, and any number of other problems. However, Microsoft stayed on top of many issues and released service packs that generally improved overall performance. In a similar vein, I’ve been enjoying the increased performance of ST over the last year while still encountering and accepting the occasional “measurement error.” I hope (expect?) the ST team will continue the platform’s upward trajectory and that our community will play an ongoing role in shaping the emerging IoT landscape. So, on that knowingly optimistic note, I’ll relinquish the soap box to anyone else who’d like to give their own sermon, or otherwise let the bug reporting continue.
Oh, and today’s performance has been quite good. Whatever problem I encountered earlier this week has definitely passed.
And @JDRoberts – thanks for the technical insights on how ST works under the hood. I love reading these gems.
No question, different people have different priorities as far as reliability, stability, versatility, flexibility, aesthetics.
But there’s also the question of market forces, and in particular competition.
There are a lot of security systems on the market that cost less than $1000 and don’t have a monthly service fee. Scout, Abode, Canary, Piper, etc. All have pluses and minuses. And then there are the ones that do have a monthly service fee, like frontpoint, Simplisafe, ADT, xfinity home, etc
If SmartThings wants to compete with those, it has to meet that standard for reliability. Just as a smoke detector, even when they only cost $15, has to meet a different standard of reliability then, say, a bluetooth speaker.
If SmartThings wants to compete with simple automated lighting systems under $1000, then it has to compete with the Phillips hue bridge, lutron Caseta, a Leviton lighting controller, etc.
So far, with the exception of people buying it based on the marketing materials indicating that it’s a security system, SmartThings has mostly been competing with low end full-featured automation systems like Insteon, Homeseer, Vera, zipato, etc. some of those are more reliable than SmartThings right now, but they all require a lot of technical expertise to use, and consequently tend to attract people who are willing to tinker in order to get more features.
That’s not a terrible business decision. Insteon is profitable and I think so is vera. But it’s a very small market.
Simplisafe does charge monthly fees ranging from $14.99 to $24.99.
Bingo, we have another @JDRoberts winner.
This is a consumer device not a laboratory, with a staff of engineering experts to both manage it and interpret the results it produces.
Not digging at you @GuyInATie, just point out that while I love to tinker and can handle ST, as I said elsewhere, I tinker to reap benefits, and am willing to tinker to make improvements and evolve the IoT in my life. I have my limits of how much tinkering I will do to keep a self destructing mess of a system from sinking itself. As you said, each has their limits. But as JD points out, your limits, my limits, everyone’s limits changes with the market.
Objectively, ST is not consumer reliable. When someone can deliver that, tolerance for shenanigans will take a decidedly downward dive.
Fixed above, thanks.