False fire alarm


#1

Not to pile on here but to register my problems, one of my First Alerts gave an alert which I received at work- first time it ever went off, so I rushed home 45 minutes to find nothing wrong… battery good. This and lights refusing to turn on/off at random, hub connecting and disconnecting has left me no option to turn off the hub until things are sorted out. Too bad because things had been fine for quite a while.


#2

Frustrating for sure, but don’t shoot the messenger :slight_smile: – ST is just passing along the report from the First Alert device.

If you have a dusty environment, smoke/CO detectors do benefit from occasional gentle cleaning with compressed air, as dust can affect the sensors.

I used to have the first generation Nest Protects, which had lots of false alarms (one while I was on vacation, so I called a neighbor to ask whether our house was on fire). I also wonder how many false alarms happened with non-connected devices and we just never knew about them if nobody was home.


(too-much-coffeeman) #3

I’ve seen false alarms as well, but from motion detectors. I’m convined that the motion detector did NOT trigger an alarm, rather Smartthings originated the false alarm. I say that in the context of other random strange issues.

I read a story yesterday how Smartthings stability problems were a natural side effect of users testing the limits of complexity, with huge control networks. My 2 cents worth is that is BS and a cop-out, as the stability problems I am seeing are NOT caused by a complex configuration. Controlling 4 light bulbs with two smart phones and two remotes was far below the threshold of “complex” defined in that story.

The good news for me is that I figured out how to connect my “broken” lightbulb problems. I use the Cree Connected bulbs. And I was told by several people that there were problems with Cree. However I found quite clearly that the problem was within Smartthings, and the fix was blindlingly simple, despite lacking that warm fuzzy feeling of a solid solution. All I had to do to “fix” a stuck Cree bulb was change the name by one character and save the changes inside of Smartthings. That change does not send anything to the bulb, but suddenly the bulb is working again. The point being that all of the people that put the blame on Cree were mistaken, or in denial.

Ever since day 1, I have seen problems with smart things keeping state with the objects it controls. I tried Wemo devices at first, and then the problems were blamed on Wemo, even though I noticed and reported that some parts of smartthings UI showed the correct state when other parts of the same app did not reflect the correct state. The bad state issue is less pronounced now, but it appears to me that it is still there, causing the problems I am seeing with the Cree bulbs. At first the fix was to remove the affected device, give the device a factory reset, and when it started working again, that was somehow proof that it was the devices fault. But I work in automation and software QA, I like to isolate the fix to the one thing… Turns out the “one thing” was simply saving a configuration change on the smartthings hub, without touching the object that is acting broken. It is not a range problem, the most problematic devices are physically near the hub, but not so close to assume a “hot” RF issue.

But the point is that the false alarms are just as likely to be the same bug I have been seeing elsewhere. I have two smartthings locations, and in both locations I see some mild instabilities. The Cree Bulbs are the worst because they affect someone other than myself… My wife. If the wife isn’t happy, no one is happy. But False alarms from motion sensors is something I am seeing with equal regularity. At my office I have three motion detectors, and three thermostats. The thermostats operate on their own, so loosing state there isn’t even visible without close inspection… But false motion detector signals are frequent enough that I never installed the alarm/strobe unit, and I have stopped driving to the office at 3am to see who is breaking in.

I am a smartthings fan, and I still recommend it to my friends. However I am also eagerly awaiting much needed improvement, so that my simple configurations stop acting like I’ve done something too complex for smartthings to handle.

I have some applications I was planning to implement in Smartthings, but due to the issues, I gone forward with implementing those in my own IOT builds. I can integrate those into smartthings or ifttt later, but for today I would rather flip the bits myself, as I don’t want smartthings turning my smoker full of meat into a fireball on my back porch because it lost state with the fan controller stuck in the ON position. A brisket can be a flavorful masterpiece, or 20 pounds of napalm, all it takes is a single stuck bit to cause your 220 degree bbq pit to turn into a 2500 degree blacksmith forge… THAT can set something on fire and potentially kill people.

But the good news is smartthings is making me think more about fail-safes and reliability. I don’t want my own controller turning my smoker into a fireball either. I’m going to have bugs too, everything has bugs… I have already seen industrial PID’s fail, so Smartthings is not alone in being buggy. Thankfully the PID failed “off” rather than “on”.

I’m thinking about perhaps deploying hadoop clusters across my controllers so that every controller in a set independently analyzes system state, and vote on who is correct, and who needs to be rebooted or shut down. So that a fire alarm or a motion detection false indication doesn’t depend on a single weak/buggy link.

I don’t know how to make smartthings as reliable as a redundant cluster of parallel systems that you poll to get the true truth from, but perhaps smartthings could develop a clustering/vote approach for those of us that want improved quality and reliability… I’d happily buy three to five hubs if it meant that reliably would approach “5 9’s”. That wouldn’t hurt sales of Smartthings hubs any… It would become a multiplier.