You would think this is basically cron under a Linux distribution of some sort. How is this still possible to be so bad?
had another random failure yesterday. This time it was different. It missed its timed task for sunrise (turn off patio lights), but without my intervention, it picked up and took care of its sundown task (turn on patio lights).
So as others, I am seeing random failures still.
I can say that as exciting as this platform is, it is clear they are still ironing out bugs, and for me this one is serious being that this is fundamental stuff here.
I could not imagine using this platform for home security that’s for damn sure!
It’s been 3+ months for me since I’ve made it through a single day of time-based events all working properly. 3 months of frustration, cold house in the morning, coming home to a dark house, and endless troubleshooting trying to find a cause. Time-based event failures seems to be 100% random. I’ve been searching for any sort of correlation or pattern but I’ve come to the conclusion that it is purely random on my end. I tried moving everything to Rule Machine, but that didn’t help. The only “solution” I can come up with is to make three instances of each time-based event, each 1 minute apart, to increase the odds of it running.
Are your times by any chance on the hour or half hour or sunset / sunrise? If so, ST fails with an overload of events at peak times. Change your times to something different, and it might work better.
Mine are a mixture of sunset/sunrise +/- a certain number of minutes and specific non-00 or 30 minute times. They all randomly fail.
For sunset/sunrise times, I would think the load should be spread out on the servers because there are hundreds of different sunset and sunrise times across different cities, unless Smartthings is using more generic times for sunset/sunrise events in different regions.
Would you mind posting a detailed description of your most common failing rule in this thread?
It is very possible that what you’re experiencing has been seen a thousand times and I’m sure we can figure it out.
I gave up on the SmartThings scheduled events a while ago. I still have a support ticket raised. I now use the IFTTT time channel to trigger SmartThings switches at key times in the day and have my routines and lights triggered on switch events. It’s the most reliable it’s ever been. The only downside is being limited to 15 minute time slots on IFTTT.
@bravenel, curious, do you know this for a fact that ST has issues with being overloaded at peak times or is it just a suspicion? Because I had the same suspicion and moved all my events to +/- 30+ minutes from sunrise/sunset and it was no better. Plus my most unreliable scheduled event is just a simple timer scheduled for 7:50am.
Just a suspicion based on fleeting observations that can’t possibly be relied on as having any validity whatsoever!!!
Who knows? And they aren’t saying anything about it…
± 30 minutes could collide with other times. I’d try something more random, just a thought.
One thing that’s nice about rule machine is you can use anything as a trigger. I tend to put related things like motions in as triggers rather than depend on times. It is the biggest single issue with the platform, and I wish we knew that there was a plan for it…
I caught it this morning completely by accident killing a scheduled job that was scheduled for 9:00:07. That time was purely coincidental, just something that happened testing something. That job was killed for running over 20 seconds. So what they’re doing is queuing things up when there is a flood of events, which I understand and doesn’t matter much. But then they kill the job as it goes stale in the queue. I bet that most of the scheduling problems arise from this one thing. Why in the world they kill a job that’s in a queue that they use to buffer high event loads is beyond me. Maybe they don’t know how to distinguish a genuinely bad app that’s in an infinite loop from one that they’ve stuck in a queue.
More details in this post:
Someone pinch me
After the Xmas break I went through the routine (which used to be daily) of fixing all my time based apps that got stuck waiting. at the beginning on November I had set up IFTTT as a back-up trigger so I hadn’t been watching ST for a while.
However since the fix at the beginning of this week not one time based smart app has failed to trigger on time. that’s about 15 time based triggers for 4 days without a failure this is some sort of ST record for me. If it keeps up I might even start to trust it…Go on fate do your worst
I was looking at one of my events that should have fired off today but didn’t. It worked yesterday but it executed at 4:46:59 just a hair before the actual due time 4:47:00. What’s weird is since it executed before the schedule, it looks like it got left behind and now it’s waiting for a day that’s already passed.
Serveral weeks since Good Night worked.
I’m going to use Tasker to issue a Good Night routine.
No response on my ticket in weeks.
Mine worked well for a couple of days after Christmas, now we’re back to working sometime, sometimes not. The application (Android) is almost unusable, keeps giving me “Sorry, an unexpected error occurred” every 2 or 3 screens, mostly when trying to access the notification screen.
Support seems totally overwhelmed, I’m doing followup every one or two weeks and keep getting the same answer for month now “Developper are looking into this”. I definitely wouldn’t want to work in the consumer service department for SmartThings these day.
My HA project has been on hold for something like 6 month now… now I’m looking at OpenHab instead.
It’s funny because my events were failing for about 3-4 days straight. For the past two days everything is working fairly well, not perfect, but much better. I think they are just switching resources around on the back-end so they allow some users to work some of the time. This way we stick around because it appears that all hope is not lost.
Hmmm, interesting. I already use Tasker for other things maybe I’ll try it for the time based stuff and run it on one of my old smartphones.
I’m already using Tasker for my presence sensing. I use SSID Near to determine when I’m home and when I’m leaving. It has been working pretty well. I am in a GPS limited area so that’s basically my best option.
I don’t know why I bother posting this, since nothing at ST ever seems to get fixed, but I’ve had two time-based events that simply did not bother to execute this morning.
OK, ST, you’ve now got three times as many time-based events from me, to increase the odds that one of them might actually work. If you’ve got capacity issues, they just got a little worse, because of your piss-poor reliability.
Nothing like random-result software. ST% really is a POS. I’m starting to think that I’ll use IFTTT for any time-based activity. It may be delayed at times, but at least their scripts actually run.
I know how you feel… I’ve had 47 failed scheduled events in the past 7 days!!!
Let me know if the multi-scheduling of tasks works for you as I also thought to do this. And my thoughts regards the extra stress on their system is the same as you!
I’m about to start building my own system using Arduino and a web/server app to replace the totally unreliable SmartThings… happy to have others join the project as well.