UPDATE: Recent SmartThings User Experience & Platform Performance

I liked the Vera Edge, but I hated the management UI and just found the whole experiencing lacking. It was, however, very reliable with very few exceptions. I used the Edge from December 2014 until September 2015 (when I got the STv2 hub). Overall, SmartThings has a much better user experience and easier / more diverse integration with other platforms…and can’t forget to mention the stellar ST community :slight_smile:

3 Likes

I googled it after reading your post and ended up on their forums, my initial reaction is I think I’d be sticking here for now. First post I read was from an ST user who jumped ship and is now thinking of paddling back to ST :speedboat:

Rule Machine is likely to be slightly more reliable, and can do anything a Routine can do, except show up in the special UI position for Routines, like widgets, etc.

2 Likes

Good point Bruce…I will give up widgets to get functionality.

You can also use Smart Rules by @obycode if you want rules and widgets all in one.

3 Likes

I think it’s important to remember that the emotion on threads like this is purely because people REALLY want SmartThings to work. Otherwise why waste the (now considerable) time following updates, questioning, waiting, hoping, etc…

There’s no denying the potential of the system, especially when you compare it other similarly priced options. What’s missing is trusted execution and reliability. Well… and more locally running services (that could help) but that’s my opinion. :wink:

23 Likes

I too appreciate the updates very much, but with that said I’ve very concerned about the long term viability of this system. When the release of the V2 hub came out and lacked the ability to run apps/scripts local, that might have been the writing on the wall. Maybe I’m wrong, but you would think if many of these SmartApps ran locally the server load would be much less.

I’ve just got done building out my system nicely and now I’m very concerned at the direction of ST’s. Executives may need to get a team together to take a very close look at what’s going on and the direction as a whole. I’m going to be watching closely on how this is dealt with not just today or this week, but long term.

Reliability is KING!!! Thanks for your efforts.

7 Likes

Wow that was an inbox explosion! Big demonstration of the need for local control to supersede everything else. I’d rather have my “automations” fire when scheduled than see whether or not my living room light is on. I’d never trust this system to security in this state or previous states.

Time to tear down everything but the core functionality and rebuild the backbone. Local first then remote functions. Hub 2 can be re-written.

7 Likes

My only wonder is why not put native SmartThings code locally on the V2 hub. SHM and other SmartThings apps should easily be ran natively on the V2 hub. Wasn’t that one of the most talked about things about it?

10 Likes

I am not a techie at all. I enjoy gadgets. Just about any type. Almost all the gadgets I get work very well. I chose Smartthings as my home automation gadget because I tried two others and didn’t like them. Smartthings had great ideas, adaptability, etc. etc. However, it seems that the “company” has fallen from grace. As I sit here, my alarm is going off every time someone opens a door or window. (I did disable the actual noisy part of the alarm.:slight_smile: The alarm is the most integral part of this system, in my opinion. If not for this community and the information it provides, I would have been long gone from Smartthings. The company appears to be making money off the good work of the users. They choose not to ensure their product works reliably before releasing it. They overcharge for door/window sensors ($40 for Smartthings sensor vs. $11 for IRIS sensor (on sale)) Not sure what I should do. Disappointed.

1 Like

Thank you for brining this up! I 100% agree with the sentiment. We want the passion, and enjoy interacting with you all, no matter what the circumstance. [quote=“bkdrake170, post:33, topic:42547”]
If not for this community and the information it provides, I would have been long gone from Smartthings.
[/quote]

Yep, our community is really great!

7 Likes

Greetings, all. Thank you all for the feedback. I know I’ve heard the kind of thing I’m about to say many times myself, but it is true: I hate when technology I rely on doesn’t work and even more so, when it doesn’t work in what looks like the most avoidable ways. I’m also not often convinced when someone from the company comes on a forum and says they hate it, too and are working on it. But, I hate it, too and we’re working on it.

So, in answer to the basic “how could this happen”:

In all honesty, it wasn’t one specific change or incident that caused this. We are constantly working on the SmartThings ecosystem. We are a constantly growing and evolving platform undergoing a lot of improvements, changes, and userbase growth. We’re still trying to identify the root cause.

From a technical perspective: There was no one leading indicator that implied the issues we saw. We rely heavily on a Cassandra database just beneath a hybrid proprietary/opensource ORM model that crosses several bridges between the groovy-ish code of the SmartApp back to our application engine and data storage. We aren’t sure yet if there was one particular trigger or several contributing factors. Some factors that might be an issue: Our userbase and device growth might have simply pushed us over a hockey-stick threshold, or maybe our work on improving scheduling and routines might have caused additional load, or maybe upcoming feature work, or potentially degradation of our Cassandra cluster as usage patterns change. It’s possible we were living just under a threshold and now we are just above. That’s one of the avenues of focus as well and if we find that is true, we’ll immediately begin work to force us well below that threshold. We monitor compactions, queries, response times, cpu loads, usage patterns and a ton of other metrics which didn’t imply failure. Does that mean we need more metrics and more granularity in monitoring? Yes, probably, that’s also something we’re working on.

We currently believe the strongest smell is from this data access layer that we might be able to pare down and customize better for our particular use. We noticed a spike in CPU load that appeared abnormal but not threatening followed by degradation of our high percentile roundtrip times (the outliers got worse) which caused us to start digging. We are tuning, epxanding, and digging still. As our previous note indicated we have implemented several immediate, temporary improvements and are stepping up mid-term infrastructure improvements that were already underway. We are also still digging for the root cause. It drives me crazy that we still can’t point to it.

I wish we had found it was 100% caused by load as we would have gotten a slow decline that would have had us expanding the cluster immediately and preventing the degradtion. I wish we had found it was 100% caused by a bug, we would have found it and fixed. I wish this was 100% caused by any one thing. We might find a smoking gun, but we might not.

We value all of this feedback and I will not excuse any failures in the platform. Our best answer to failures and complaints will be to make things better.

37 Likes

Smart home monitor page is crazy slow to load and I can’t get it to leave “Arm (Stay)” mode. Not sure if the issue is supposed to be over or not. My alarm keeps triggering and sirens keep waking the babies. Hope I don’t get arrested!

1 Like

Thank you for sharing. Basically you are telling me there is no extra load to put in place while dealing with the current issue. That’s not cool.

1 Like

Agreed! Spin up a few VM’s or do a hot add. Tell us your hypervisor/hosting provider are we can give you instructions: https://pubs.vmware.com/vsphere-50/index.jsp?topic=%2Fcom.vmware.vsphere.vm_admin.doc_50%2FGUID-73B63A2C-96D4-4C14-80A3-3A698DC48F06.html

3 Likes

Thanks for the note. Much appreciated. As I have said, reliability which includes good response times should be a priority. It’s that simple for me, but I know it’s not that simple on that end.

1 Like

Thanks for the update. I have now removed SHM completely waiting for a positive news to put it back. I moved from iSmartAlarm after a year using it and it was rock solid when comes to reliability. It is a closed system, but worked.

I suggest all critical functions (SHM is the most critical) should be handled locally.

As one of the people who has been most critical of SmartThings (partially related to some of the inside knowledge I was privy to while briefly at Samsung after the acquisition), let me say that you coming out and explaining this @bakken7 goes a long way toward beginning to repair some of the trust of those of us who would absolutely be your biggest supporters.

As another response in this thread mentioned, the reason that you have passion (and emotion) from a lot of us is that we feel we’ve invested not just our money, but our time/effort/blood/sweat/tears to implement and push this platform to places it probably wasn’t originally designed to go. :slight_smile:

I will only say that you all have to find a cleaner way of doing devops that doesn’t necessitate these kinds of heroic efforts every time something gets tweaked. Hopefully, you guys know that, and thanks for continuing to pound away and work at this.

8 Likes

Just be aware that not everyone has the same level of criticality of components… I could care less about SHM… I just want my lights and motion sensors and rules to work reliably so I don’t wind up in divorce court. :slight_smile:

25 Likes

In instances where run-away loads are happening “just throwing power at it” isn’t going to help and I’m sure they’ve already tried this.