This morning my smartthings apps failed to trigger on even the most basic things (like motion in a room turning on a light) for about an hour. I posted in the general forum with details and asked if anyone else had this issue (http://build.smartthings.com/forums/topic/apps-not-responding-this-morning/).
I checked the facebook page, the general forum, and the blog, but didn’t see any info on this from smartthings.
There’s really no way for me to know what happened, but after about an hour of unresponsiveness, all my apps fired on the triggers that had been happening for the last hour. It feels like the actions all got queued up and just never sent.
It feels very much like a congestion/glitch issue with the smartthings cloud, which prompts me to ask - is there any news on local hub-based control? Or any kind of health-monitoring/high-availability feature on the radar so that I can troubleshoot better?
I was having major issues yesterday installing applications as well. I am curious how the local control works right now. If I unplug the hub from the internet, will everything come crashing down? I thought there was some sort of local cache or failsafe?
There was an issue this morning with some queues backing up and it slowed performance. We are improving that infrastructure daily. There will always be some functionality that will require a connection to the internet. And while we do have plans to bring some of the controls into the hub itself, these features are not in the immediate future. For that matter, we do not know that the existing hubs could handle that, in any case. It would likely be that we would have to release a version of the hub with a more powerful processor.
I would think that a secondary (optional) device to represent a local cloud woudl be smarter. Have a primary IP for the SmartThings cloud, and a secondary to an (i.e.) RaspberryPi which can do the processing. Then you can have an upsell, and not “betray” your original clientele by requiring a replacement hub/device. (Unless you would allow a trade-in at small cost, and then you guys sell the hub as a refurb)
+1 I love this idea. I could have a service running on my 24/7 server that could act as a back-up just in case the Internet link is severed. You never know what can happen, especially with a consumer grade connection. Right now my SmartThings is responsible for code management on my Kwikset deadbolt, and auto lock/unlock based on proximity. My roommates do not concern themselves with locking the door when they leave, as it’s been 99% reliable.
I gave up on ST for this very reason. “Cloud first” that ST had adopted is all fine and dandy until you got hit with Internet blackout or server congestion. Heck, I don’t want power outage on the East coast preventing me from opening my door lock in California. Even worse, what if a command stuck in a queue would trigger an hour later and opened my front door when I’ve already left my house?
I also believe @ben_edwards is right and existing hub will never be able to handle local control. ST shot themselves in a foot by choosing PIC32-based system instead of embedded Linux, like everybody else does. They may have saved few pennies, but now they’re stuck with a rigid and underpowered platform. Looks like “Cloud first” is going to be “Cloud only” for a while.
@darrylb: Don’t get me wrong, there are many things I like about ST. Yes, I do believe their “Cloud first/only” approach is a bit naive and for most real-life situations, particularly alarm and security applications, you absolutely need to have local control. It doesn’t do you any good if your siren goes off an hour later after your house’ve been broken into. But the software they’re building is quite powerful and open. I’m waiting for it to mature.
I just started looking into SmartThings, and this is my biggest question about it. I have a server at home. It’s an 8 core AMD. It’s not the fastest thing on earth, but it is easily fast enough to support one more service which serves as command and control for the hub. The hub itself doesn’t need to be that smart, and my server doesn’t need to be augmented by cloud processing, and I’d rather develop using VS rather than trying to develop on a website. I’ve read the introduction and what we believe sorts of docs and I don’t really understand what they are trying to say. I already have an IDE I like and don’t want a different one, and I already have a server. I don’t understand where the cloud comes in when I’m attempting to control my house from some sort of device (PC, tablet, phone) which is in my house via a server which is in my house. When will I be able to configure SmartThings to be local network only?
My guess for the answer to your last question (“When will I be able to configure SmartThings to be local network only?”) is never.
I would suggest that you check out the Vera Lite or Vera 3. They can be configured to be local network only. When I want to access my Vera Lite remotely, I do so through a VPN connection.
That being said I still greatly prefer the SmartThings. I find the user experience and support from the company to be far superior to the Vera.
@johncuyle: the short answer is “probably never”. SmartThings is built on the premise of “Cloud first” (their own words). This definitely makes a lot for things much easier (for developers that is). As for the users, some people prefer simplicity of setting it up and ability to control it outside of the local network - there’s no need to mess with the router settings to open incoming ports, configure dynamic DNS, etc. If you don’t care about latencies (sometimes several seconds) or that your system may become useless because either your internet connection is down or the SmartThings server is not responding, then it may be what you need. If, on the other hand, you require fast response and near 100% availability, it may not. I have several home automation gadgets, for example 3M WiFi thermostat and TCP Connected lights, both of which have cloud AND local control. And the cloud is optional in both cases, i.e they work just fine if I turn the Internet router off. So it’s not impossible to have both, it’s just ST for whatever reason decided not to.
I read that in their “What we believe” page. Except that none of that is true. Software development using a local machine and, at most complex, remote debug across a LAN is vastly easier than developing and debugging a cloud service. Developing an application using an IDE you know well and a language you know well is easier than typing text into a website. I already have dynamic DNS set up and already have VPN set up. There’s zero overhead to controlling a local-only solution remotely. I do care about latency when I turn on the lights and I do want my lights to work when the internet is down. I see that they explained their rationale, but it makes no sense and their assertions are easily falsifiable. I was hoping that it was just marketing fluff until they have a local solution in place.
I agree with you that user support and experience smartthings is superior to Vera, but having a controller that does not require internet access may sometimes break a deal. I, personally, would love to have both options available and will gladly buy two. Presently I am trying to integrate Vera with Smartthiongs for this very reason - no internet required in case of emergencies.
@johncuyle: I’m definitely with you on the home-server as controller, and hub still as what it is (a go-between). @ben_edwards - what does Smartthings think of that kind of future, for those of us who are savvy enough to do such a thing?
Anyone else have more problems this morning? I’ve been turning stuff on and off manually all morning, only to have things turn back on (due to delayed triggering, I assume) minutes later.
none of my smartapps ran this morning, either at their scheduled time, or if I manually ran them. but when I manually changed the temperature of my thermostat in the ST app, it did change the temperature on the thermostat.
Stuff is still glitchy this morning. The general forums had people respond to my post saying they’ve been having similar problems for days. Can you give us any more info on what is going on, or just when it will be resolved?
This is the way ST works. Sometimes it does, sometimes it doesn’t. In my opinion it’s still in development and you just have to accept it with its flaws in the hopes that it will get better. Personally I think they’re making a huge mistake with their huge PR marketing push when the product is not functioning 100%. I think that should definitely be taken care of first.
@jeremywhittaker: yeah, you nailed it! That’s how all start-ups work. Consider this: ST was originally funded by Kickstarter and some angels, raising around $1.5M total. Last fall they took around $12M Series A funding from a bunch of VC’s, with total valuation around $15M. That’s x10 ROI in one year! Sure, VC’s don’t work for peanuts, they expect x10 return YoY. Series A funding usually runs for 12-18 month, so we’re talking about $150M valuation for the Series B. Now, how do you value a company like ST? There’s no way they make any money selling hardware (not yet at least) and they don’t charge for the service either. So the only way to value them right now is by the subscriber base grows. It’s all about numbers now. That’s why all the CES publicity, advertising and marketing hype. And when you double, triple, quadruple number of users, cracks will surely appear in their cloud infrastructure, this is simply inevitable. I’m sure ST will do their best to patch them as soon as they can, but it could be a bumpy ride for a while. So yeah, we’re all beta-testers here, like it or not.
