Is it time to completely stop using routines and smart lighting? (rule machine)

So, like many others, I’ve had massive issues with routines not firing, (time and presence, and any really,) reliably. I’ve also had pretty crappy luck with Smart Lighting. In fact, the locally-installed automations were LESS reliable than the cloud-based ones.

I’ve moved almost all my lighting automation to Rule Machine now. I assume the default reaction is “uh–dumb, that’s not local at all.” – and at first I would have agreed with you. Now, when I open my closet door, the light is on before it’s even fully visible. Instantaneous. Everything I have set up using Rule Machine is working just like this. I have also added additional logic, like an “Auto Lights” virtual switch to control each section of my house… as a variable basically. Of course this is not really possible with Smart Lighting, and I do not fault it for that.

I do wonder, how it is that cloud-based, user-provided smartapps are running complex lighting automations more reliably and quicker than the OEM solution that runs natively, locally.

So now, with months and months of issues with my routines not firing properly and playing tag with ST support, I am considering doing the same thing with Routines–and moving them to Rule Machine. My greatest fear is that somehow I will overload the RM framework–or that I will lose these all during an upgrade in the future. I really, really, wish there was a way to backup these automations. Perhaps the smartapp could have an import/export option for google drive, or something. That would be awesome.

Does anyone know if I can add a second version of Rule Machine, and call it “RM Routines,” or something? Not sure how the child smartapps communicate to the parent–if the name matters in this code. I hope this doesn’t sound like pure whining—because I’m not trying to. I just want to use whatever works best. So far, writing this all out using RM is working much, much better than using Smart Lighting, and I’m confused as to why.

1 Like

I’ve moved over almost ask of my lighting tasks to RM. I have a few still on the smart lighting app. But those are things like the laundry door turning on the lights and real basic stuff.

I don’t use any ST routines, they have ask been migrated to RM.

I don’t use modes… I used them for a bit but too many failures drive me away. I just use virtual switches in RM to act as modes and use a rule to turn them off and on… They are the triggers for my routines that are mode based.

It works great!

So far, I haven’t found a need that I couldn’t complete with rule machine.

I actually only have the following smart apps installed:

Amazon echo
Battery monitor
Button controller
Hue connect
Rule Machine
smart lighting

At last count I had 65 devices connected and it running pretty darn good… (I would say really reliably but @jdroberts will call me out and publicly shame me lol!).


I’m considering moving away from modes as well. Sure would be nice if we could just set global variables instead. Virtual switches might be my answer as well, but still messy. Modes were nice, because they limit even RM based on a variable I didn’t have to tediously enter into the RM smartapps. Once again, if I could somehow create a complicated rule from a template and input from Google drive, etc, that would totally rock.

Curious, for actions based on presence, do you do a trigger, or rule?

I am in the process of converting to using virtual switches to hold states (time-of-day, day-of-week, presence). RM is being used to manage those and make sure all of the switches are in the proper state, and also to set the (more-limited) modes. Also using RM to drive state-changes and where possible using RM actions to make things go.

In some cases I am using Routines and other SmartApps to make things happen. But for the most part I’m leaning towards RM, as the main driver; it seems to work pretty well and @bravenel is very responsive (and doesn’t introduce too many new bugs) in his efforts to implement new functionality.

It seems totally illogical that a cloud-based user smartapp is working far better than a locally-running ST-designed solution. I’m having a hard time wrapping my mind around that.

1 Like

Well, it’s all about semantics here:

Reliable means consistently good in quality or performance; able to be trusted
Stable means not likely to change or fail

Do you really “trust” ST now after couple of weeks that they won’t bring your set up down, again?

I’ve been in your shoes, and virtual switches turned out to be less stable than modes. The sad part is that ST proved that neither methods are reliable. So I am just waiting for better times, while cruising with a set up that is of minimal complexity and taking advantage of the local processing that has proved somewhat stable for me in the past two months!

Yeah there’s no reason that a virtual switch can’t fail to change as well. Realistically there’s no difference – just a dang variable.

Really, I’d rather NOT move all my mode-logic over to Rule Machine. That’s going to be a nightmare. I have home/away modes for 5 timeframes of the day.

Been there done that! Pulled my hair out and got a high and tight hair cut down to just two modes, 4 routines, a bunch of RM rules and several custom apps. Not what I want from my HA set up but at least I have no hair left to pull when things go bunkers!

Do I trust ST to work the first time, every time, without question? Not a chance. ST has proven they can not handle the server load.

I do however believe there is a level of reliability, not incredibly high, but it is there. To come to that conclusion I compare the system to it’s direct competitors (that I have experience with).

It can not be compared Phillips hue. PH only does lighting, and they do it incredibly well.

It can be compared to Wink. I believe Wink has a much higher level of reliability, as far as the device will eventually do what it is supposed to do.

Overall, I am happy with ST. But, I am not loyal to the brand. If someone else comes out with a stable and fully reliable system, then I’m jumping ship.

I’m really hoping that ST recognizes their lack of server robustness and they are able to bring the needed resources online.

And I will say this, over the last couple of months my system had become more complex and it had been fairly reliable (but it is high maintenance and I’m getting tired of that game) until this week.

The modes and routines were bad a couple of months ago. RM has not been affected, until now.
Monday I had a handful of rules missing variables, so of course they won’t fire. This morning is the first time my good morning rule has failed to fire since I think early November. No idea why, other than it is a timed event.

This morning I could not even get a timed rule to allow me to talk done and exit. I kept getting the red banner, which of course kept the rule from being evaluated.

I also realized I have no way to manually fire any of my timed rules, which will be fixed today. I’m rethinking the logic on them and I am sure I’ve come up with something.

Stable? NO when based on an off the shelf consumer system.
Stable? Yes, when compared to Sept-Nov 2015.

1 Like

Remember on the v1 hub when we had the Dashboard? Things were nearly 100% reliable for me.

So I just moved over my routines (leaving some for manual-trigger,) and what’s immediately pleasant to me is that my 20 routines required for mode-control are now 10RM, thanks to true/false actions. I could probably narrow it down further, but then troubleshooting will become a nightmare.

I’m using both routines and RM to try and get at least some reliability. Both are setup with the same time and settings and if I’m lucky one of them will properly run everything. I also ditched Smart Lighting and use RM for all of those features. RM had been working fine for the lighting, until 2 days ago.

And now this week the servers are deleting variables out of rules in RM and now timed event rules aren’t firing right…

Coincidence? ? ? ?

Hmmmmm… Maybe RM is making someone look bad? ? ? Lol

1 Like

There were ST updates that impacted all time-based solutions within the ecosystem.

Again, a backup/restore mechanism for RM would really help with ALL of this.

The best and easiest back up is the one that Bruce @bravenel is using. Screenshots, screenshots and some more screenshots worth of 20 char screenshots, that is!

1 Like

I still vote import/export!


maybe if we ask nice, @Mike_Maxwell would find a way to write a report to export the rule to an xml file and then to import it to google drive? Is not far fetched but I know he has plenty on his hands…

I imagine the functionality would work by setting the GD credentials on a main preferences page, then at the bottom of each new rule, there’s a backup and save button. The ability to restore from full deletion would be unbelievable. This would also potentially give us a mechanism to design rules in a text editor. Big win!

I already have a complete parent child demo app that can backup the child settings to the parent and create a new child based on the settings that were backed up from other children.
I’m working with @pstuart on creating a demo web service to get these settings off site.
Last Patrick and I talked we were considering working on the web service aspects this week on LCF

So it is technically doable, however not sure if it’s feasible with RM, my demo app has 5 or 6 inputs that get restored. I’ve not counted up all the potential inputs that RM has, but it’s a huge number compared to the typical app.


It would be pretty cool if ST would allow us to back up/modify smartapps from a browser as well. :smile: