Long Running Randomness

[larryllix]

On 10/31/2024 at 10:20 AM, CoolToys said:

@larryllix, I understand where ISY came from, I started out with a 99 when Leviton purchased HAI and made a mess of a great product before discontinuing it.  That said, the program you said will never run works just fine.  In both versions, I have tested each and they both work exactly the same.  While I trust you have a deeper knowledge of ISY than I do, I do like to test things before I take them as gospel.  That Late_Arrival program is at least 5 years old and hasn't failed once.  Yes I have searched for that variable and it is in two places.  "Auto_Late_Arrival" where it is set to "1" and "Auto_Late_Arrival_Off" where it is set to "0".  

The same methodology is used for "Late_Night_Snack" as well, basically the opposite of arriving late, it is getting up after bedtime and leaving a bedroom.  It lights the same path backwards.  Again with "sSnack" at the beginning of the code, the lights go out on time in order.  So again I am confused by this atomic thing since the code works just like old BASIC.

The only one doing random things is the Bedtime program, and it hasn't gone wonky in 10 days now.  It ran very slowly last night, taking about twice as long as normal.   The place I am watching to see if I have a code error is if it is "Night" and the house isn't occupied and it becomes occupied.  The value of "sOccupied", "sGuest" and "sNight" changes the lighting in the house. If no one is home, only the lights which are visible through windows on the outside of the house come on.  If there are guests, a couple more lights in that section.  When we come home and the house was empty a couple more lights come on and the states all adjust. 

Last night we came home late, logged in from the driveway and verified the house was in Night Empty mode, and I watched it changed to Occupied mode when I opened the garage door.  Bedtime worked fine, so I am still stuck as to why on occasion it acts like the alarm is set or someone is moving around the house downstairs.  Disabling or Removing the ELK poly may be the next step.   

The other thing I want to dig deeper into is the autonomy of the eisy with Poly's.  Is the eisy dependent on an internet connection?  Once I pay for a poly is the eisy autonomous or connected? If it is connected, why not just use Alexa or Siri which both control my Matter blinds just fine as well? I started with HAI and moved to the ISY because I wanted to build an "Autonomous" smart house, not a "Connected" house dependent on an outside service.  One other possibility is that the internet connection was not working when the "bedtime" program went wonky.  All of this goes back to my original thread about wanting the logs to see what programs run when.

 

OK. I ran some testing to prove this out,  one more time. The trigger logic is just as I have stated on my polISY v5.8.4. ISY program logic hasn't changed on my system.

There must be something not right about your program logic, or eISY.

When I run this program the result is a "2" count.
This tells me logically that  $sTestTrigger = $cTrue  and $TestTriggerProgress lines  are atomic, until the next Wait is encountered, which then allows the system (by surrendering it's time slice),  causing the If section to re-evaluated, stopping the running of the True section and logically running the False section.

The following True section lines (after the variable change and next Wait line) will never be executed.

TestTaskSwitch - [ID 010D][Parent 0001]

If
        $sTestTrigger is $cFALSE
 
Then
        $sTestTriggerProgress  = 0
        Wait  5 seconds
        $sTestTriggerProgress  = 1
        Wait  5 seconds
        $sTestTrigger  = $cTRUE
        $sTestTriggerProgress  = 2
        Wait  5 seconds
        $sTestTriggerProgress  = 3
        Wait  5 seconds
        $sTestTriggerProgress  = 4
        Wait  5 seconds
        $sTestTriggerProgress  = 5
        Wait  5 seconds
        $sTestTriggerProgress  = 6

 
Else
   - No Actions - (To add one, press 'Action')

However, when I run this program below, the end count, $sTestTriggerProgress, becomes "6".
This tells me the time slice has not been surrendered until the end of the program True section and demonstrates the difference to the interrupted program above.

TestTaskSwitch - [ID 010D][Parent 0001]

If
        $sTestTrigger is $cFALSE
 
Then
        $sTestTriggerProgress  = 0
        Wait  5 seconds
        $sTestTriggerProgress  = 1
        Wait  5 seconds
        $sTestTriggerProgress  = 2
        Wait  5 seconds
        $sTestTriggerProgress  = 3
        Wait  5 seconds
        $sTestTriggerProgress  = 4
        Wait  5 seconds
        $sTestTriggerProgress  = 5
        Wait  5 seconds
        $sTestTriggerProgress  = 6
        $sTestTrigger  = $cTRUE
 
Else
   - No Actions - (To add one, press 'Action')

 

[dbwarner5]

@larryllix great test program. Clearly verifies that the ISY rules work that way, as we all expected it should. 

Means something is corrupt. 

@CoolToys  Sorry, but I cant remember, did you ever delete the program and rewrite it from scratch? Maybe in a different folder even. just to test if it's the program or the system? 

[CoolToys]

@larryllix, I agree with @dbwarner5, great test, and I get a similar result when I used "0" and "1" rather than True/False.  So I used the same methodology for my existing program, and all three lights went off, but when I changed to 5 seconds, then the lights stopped where I moved my variable "sAtomic_Test".   Thank you, and I have moved my resets to the bottom on all three programs that use this, and I will try and figure out what else could have been turning the lights off 30 minutes later.  

And @dbwarner5, no I did not delete and rewrite the Bedtime program and since it has worked for 10 days now, I am not sure how I will know if that is the fix unless it fails again, but I will do that just in case.  

An interesting side note.  All of my notifications related to the ELK Alarm stopped working on the 20th.  That is about the same as the last time it went wonky.  So not only am I going to rewrite the program and stuff it in a folder, I am going to look at all of the places I use the ELK poly.  Thank you both, for the time and I will post here if I figure out what the issue is.  The Elk module says it is working and needs an update so clearly I have some work to do.

Update, the "notification" module wasn't working, so a quick restart and notifications work.  So... I wrote the new program but left it disabled so I can see what happens tonight.

Question then comes up with the "Bedtime" due to it's size, the momentary "off" starts the program, but the state of that switch never turns "on", so this is probably where I have some things really jacked up.  There is a 6 button keypad on each side of the bed, one old insteon, one new, so one appears as a 5 button, one as a six.  Should build a scene that activates when "bedtime" no longer applies to set both buttons to on or is that way over thinking the issue?

Update 2 - I started testing every single scene and program one by one, and found that three of the "scenes" do not work at all.  They are three way lights with an insteon dimmer at each end. Three others work. Turning on the master works.  The slave does not set the master, and the master does not turn on the slave.  In the Admin Panel, when I try to turn the scenes on and off, nothing happens.  I deleted the scene and rebuilt each of them, so there is clearly something that didn't migrate right from the ISY or got corrupt along the way. 

Edited November 2 by CoolToys
update one item fixed. question added

[larryllix]

5 hours ago, CoolToys said:

@larryllix, I agree with @dbwarner5, great test, and I get a similar result when I used "0" and "1" rather than True/False.  So I used the same methodology for my existing program, and all three lights went off, but when I changed to 5 seconds, then the lights stopped where I moved my variable "sAtomic_Test".   Thank you, and I have moved my resets to the bottom on all three programs that use this, and I will try and figure out what else could have been turning the lights off 30 minutes later.  

And @dbwarner5, no I did not delete and rewrite the Bedtime program and since it has worked for 10 days now, I am not sure how I will know if that is the fix unless it fails again, but I will do that just in case.  

An interesting side note.  All of my notifications related to the ELK Alarm stopped working on the 20th.  That is about the same as the last time it went wonky.  So not only am I going to rewrite the program and stuff it in a folder, I am going to look at all of the places I use the ELK poly.  Thank you both, for the time and I will post here if I figure out what the issue is.  The Elk module says it is working and needs an update so clearly I have some work to do.

Update, the "notification" module wasn't working, so a quick restart and notifications work.  So... I wrote the new program but left it disabled so I can see what happens tonight.

Question then comes up with the "Bedtime" due to it's size, the momentary "off" starts the program, but the state of that switch never turns "on", so this is probably where I have some things really jacked up.  There is a 6 button keypad on each side of the bed, one old insteon, one new, so one appears as a 5 button, one as a six.  Should build a scene that activates when "bedtime" no longer applies to set both buttons to on or is that way over thinking the issue?

Update 2 - I started testing every single scene and program one by one, and found that three of the "scenes" do not work at all.  They are three way lights with an insteon dimmer at each end. Three others work. Turning on the master works.  The slave does not set the master, and the master does not turn on the slave.  In the Admin Panel, when I try to turn the scenes on and off, nothing happens.  I deleted the scene and rebuilt each of them, so there is clearly something that didn't migrate right from the ISY or got corrupt along the way. 

On the rewrite note. I have found in the past that programs can get out-of-sync with their tokenized codes at some weird times (usually transferring systems) . I found the easy fix for that was to click/select each line of program code so that the parameters are all found in the edit section below and then click on update, so the ISY code compiler regrabs all the token codes and refreshes them. Do all lines of code and then click Save for the program. It's an easy to do trial.
Of course you may want to disable and then re-enable your program before saving, also. It could be magically disabled and may explain the "works anyway".

IIRC you posted you did a search for any other usages of your trigger variable. It really seems like there is another hidden program changing your variable on you.

[CoolToys]

@larryllix, Last night using two computers in the bedroom to watch the logs, variables and states, I may have found it, and figured out why it is "random".  Yes as both you and @dbwarner5 noted, something was corrupt.  Not understanding the "token" it makes more sense now.

Several of the scenes didn't work and in program code for some reason I changed directly to the device instead of fixing the scene then, but it worked so I didn't think about it.  After fixing scenes (deleting and rebuilding the three that didn't work) and replacing the direct commands to scene commands in the Program "then" and "else" lines, something different happened.  There was a motion sensor at the top of the stairs with a low battery that was randomly sending "On".  When Bedtime was pressed, there are two different programs triggered by the motion sensor.  That was stopping the bedtime run and setting the house back to "occupied".  Why it took as long as an hour is still a mystery, and why I say I "may" have found it.

This is a flaw in my logic where the Bedtime was setting the house to inactive, and the stairwell was resetting to active before Bedtime could complete. I added a statement to the "occupied" program to check if "$sBedtime" was = "1". Scene fixed new, battery, variable state changes moved. and so far so good.  Fingers crossed since I hit the road again and my wife has early days all week.  

Everything done last 48 hours:
1. Elk Poly upgraded and rebooted
2. Notifications poly rebooted.
3. Three Scenes that were for three way switches, two switches one load, all deleted and rebuilt with a new name
4. Device names in programs replaced with scene names
5. Battery in motion sensor replaced.
6. Added "sBedtime=0" to Occupied check program
7. eisy backed up and restarted.

Thank you both again for all the ideas and things to check. 

 

[CoolToys]

5th night update, 
So far so good.  Thank you again for explaining the tokens so I would verify every scene I just assumed worked.

 

[larryllix]

12 hours ago, CoolToys said:

5th night update, 
So far so good.  Thank you again for explaining the tokens so I would verify every scene I just assumed worked.

 

Been awhile since my MS 1s were not packed away but this brings back memories of them (I had 12) producing random On_s when the batteries were low.

Bad Insteon...bad!

[CoolToys]

Agreed, bad insteon.  One of the old motion sensors went out last night and my closet just flashed like a disco ball.  Ever since selling my Lutron/Crestron dealership, I have been pretty happy with insteon, but then as I look back over the years and all the problems I have had, they always came down to two things.  My programs in the iSY (HAI just worked but didn't have if then else, only if then so much harder to goof up the linear logic) or insteon.  Now that both Thermostats have gone totally wonky, I am looking for all new lighting and HVAC control.   The hard part is that no one has the outlets (not wall warts, but real outlets) and light switches except insteon.  I am as you say "committed" to some extent.  

New insteon doesn't have thermostats, so today, I am trying to find a couple of matter thermostats that I can control with the AC.  I would love to stay totally autonomous but the AC won't run my matter greywind blinds that Apple/Amazon and Google all find and control perfectly.  My M1 can do basic lighting control, but not even as good as the Insteon hub.  I don't want a thermostat that tells any of the big three I am home or not, so nest and company are off the list.

 

 

Edited November 7 by CoolToys
Thermostat failure addition.

[larryllix]

4 hours ago, CoolToys said:

Agreed, bad insteon.  One of the old motion sensors went out last night and my closet just flashed like a disco ball.  Ever since selling my Lutron/Crestron dealership, I have been pretty happy with insteon, but then as I look back over the years and all the problems I have had, they always came down to two things.  My programs in the iSY (HAI just worked but didn't have if then else, only if then so much harder to goof up the linear logic) or insteon.  Now that both Thermostats have gone totally wonky, I am looking for all new lighting and HVAC control.   The hard part is that no one has the outlets (not wall warts, but real outlets) and light switches except insteon.  I am as you say "committed" to some extent.  

New insteon doesn't have thermostats, so today, I am trying to find a couple of matter thermostats that I can control with the AC.  I would love to stay totally autonomous but the AC won't run my matter greywind blinds that Apple/Amazon and Google all find and control perfectly.  My M1 can do basic lighting control, but not even as good as the Insteon hub.  I don't want a thermostat that tells any of the big three I am home or not, so nest and company are off the list.

 

 

Yeah. I am a dedicated Insteon person but the old Insteon made some real garbage. I had two 2441WTH stats and one never worked from day one. UDI support pointed out that when sending a command the more recent 2441-WTH model was acknowledging the request but would never react to any commands. It became a probe only unit. Should have been 2441-WTF.

Warranty? Sure but I had to pay the shipping from Canada across the border for about $65.

The earlier model worked OK but the 3c = 5F dead-band was intolerable for us, freezing draughts at some times and heat waves of sweat the next hour. I dedicated that one to simple ventilation in my cold-cellar to control humidity.

Again...Bad Insteon!  I hope to keep hearing more responsible engineering from the new Insteon. They need to change their name to Insteon2 in they want to circumvent some of the bad taste so many users got in their mouths from the last idiots.

Still a good (possibly the best) system and I wish the new owners all the best.

Edited November 7 by larryllix

[CoolToys]

New insteon owners don't have a thermostat and and as @larryllix noted the conversion isn't usable for me either so I am looking for a larger cooler looking thermostat that isn't tied to a paid or central service.

 

[larryllix]

10 minutes ago, CoolToys said:

New insteon owners don't have a thermostat and and as @larryllix noted the conversion isn't usable for me either so I am looking for a larger cooler looking thermostat that isn't tied to a paid or central service.

 

ecobee is a cloud monitored system (not cloud dependent) and very well done NS, and device.

I have owned many different brands of stats in my life and ecobee has been the best one (I have owned three now) I have ever experienced. However, the units having Alexa installed in them have been complete failures. You can't put a microphone flat onto a wall and expect it to understand you properly. Alexa manuals also tell you that.
I also find the remote sensor idea is a complete failure. I have three remote sensors and leave them all disabled after a about five years of experimenting with them. Silly idea without a door closed sensor to change it's personality.

Edited November 7 by larryllix

[CoolToys]

@larryllix, thank you , ecobee is on the top, as I understand there is a way to disable the cloud portion.  There are a couple of others that are "matter enabled" but as I found with my Greywind/zooksmart blinds, that works great for Apple/Google/Amazon, however UDI not so much.  The other issue is the wall behind the thermostat was never finished from the original home and the shape of the ecobee means I have to finish, texture and repaint the wall, but it might be worth doing.  No real need for remote sensors, but it looks like to get the upgraded model, the sensor is part of the deal?  

Back to this thread, everything is still running ok.  Bedtime again took way longer than it used to but ran perfectly,  Maybe it realized I am getting older so it slowed down to look like me walking through the house.

The older Insteon motion sensor, is still acting up, so it is getting replaced with a new one, and last night my garage lights didn't go out and the AC said communication error.  I queried the lights and they responded and turned off.  I had that issue a while back and it was fixed by replacing the "CyberPower" brand of power strip on my work bench with APC.

[paulbates]

I've had good luck with the remote sensor that came with my ecobee. Keeps the temps better balanced without resorting to baffles or individual vent control. The 'occupied' feature had the halo effect of it turning off the suite lights off after 30 mins if left on

[CoolToys]

@larryllix and @paulbates, I ordered both a Sensi and Ecobee.  I have a two zone house that is pretty well balanced  so I don't need to play with vents or baffles either.  Why I bought one of each ($20 cheaper and delivered today on Amazon instead of direct) was to see how much data the devices pass outside of the house.  Neither is going to be connected to the Google/Alexa/Apple eco systems, and I am hopping eISY can control one of them without it checking into the host regularly.  The entire reason I have an eISY is so my home can operate autonomously, but that dream appears to be fading.  The new Crestron gear is starting to interest me.  

[paulbates]

@CoolToys Yes, I've had to change my approach to security that was previously local only. I have yolink and ecobee that each require cloud side. Each of their plugins access the devices in my house back through the cloud accounts.

One partial advantage vs local only is that I assign ecobee and yolink to an iot/guest network so they can't see my other phone/tablet/pc devices. I also "force feed" both of them NextDNS's ad/spy/malware blocking DNS addresses to limit their data collection or interaction with a nefarious host.

I also "caved" and using the Alexa on my ecobee but it's constrained the same as the ecobee. 

Edited November 10 by paulbates

[CoolToys]

@paulbates, I use a similar tactic, with two different VPN's, one for house and one for guest.  I had to port my eisy out so that UDI could help a while ago, and I am about to close it back off and see if the Poly apps keep working.  I see data going back and forth but can't see what it is.  

Sadly the Alexa multi room music feature is the reason I caved in the main living area, however I am considering another solution since it randomly stops the music quite often.

[larryllix]

I have never needed to operate my ecobee via voice control. However, I have used the webpage controls to modify settings and set it up. That takes cloud access. I am not sure how secure their comms are with the stat, but it can't be any less secure than all the VPN personnel, people are giving full control of everything in their house to.   I have set limits on my ecobee stats and a few watchdog ISY programs to monitor ridiculous settings, so the worst attack should be an attention getting cool or warm spell in the home. Ecobee stats have those limits that can be setup under password control lock.

Previously, I did setup a low security home monitoring system with arm/disarm control via Alexa voice. However, I used a double command sequence, along with a time window allowance, to make it more secure. More intelligent hacking would not be likely, for the benefits gained.

[CoolToys]

Well @larryllix and @dbwarner5 and @paulbates, the Sensi Touch Smart 2 is supposed to be Matter ready but it is only with the Samsung matter app using wifi.  Epic Fail, getting returned, which is too bad I like the look way more than the ecobee.  Still trying to get the ecobee to connect to the eISY, but giving up for tonight, regroup tomorrow.  Thanks again.

[larryllix]

I have been holding off on anything Matter. I have a few dozen Insteon plug-ins that I will never use now. I started on Z-wave buy a z700 dongle and a few plug-ins but once I saw the 1990s style interfaces incorporated into ISY and the poor quality system complaints, some of my new units have never been out of their boxes.

The Matter implementation is over Ethernet via LAN so I haven't seen any reason to make any jumps to  it until the clothes come out of the wash totally clean. Most of my lighting is multi-coloured RGBWW/CW WiFi bulbs and they work very quite well without any cloud connections at all. Of course my 18-20 bulbs are getting to be 6-8 years old now.

[CoolToys]

@larryllix, I am coming to that conclusion as well.  I really liked the idea of Matter, but as someone here posted "does it matter?"  MK and I had a conversation about having an open standard so we can make this stuff work and his response was no more standards.  Sadly every manufacturer of RF or Matter devices is catering to Amazon/Apple/Google and occasionally Samsung, we are getting left out of some great stuff.  My automatic blinds are a game changer to my sleep and home A/C usage in the summer.

All of my matter devices work great with the Alexa and Apple Home Kit ecosystems right out of the box.  I don't understand why I can poll them with the UDI/eISY but just get "unavailable". I don't want Amazon or Apple knowing when I am home opening and closing blinds, bad enough half of my house uses Alexa for music, but I think I am going to change that.  Half is already wired, so I am thinking of going wireless, non amazon and using my REQUEST Server more or building my own winmedia server since the Request can no longer stream.  I am getting tired of walking in my kitchen to ads about faucets, and my watch telling me I slept in late again.  Just for fun I split my matter devices between the Apple and Alexa systems to see how it affects ads and stuff that shows up at my house.

Quote

 

 

[larryllix]

20 hours ago, CoolToys said:

@larryllix, I am coming to that conclusion as well.  I really liked the idea of Matter, but as someone here posted "does it matter?"  MK and I had a conversation about having an open standard so we can make this stuff work and his response was no more standards.  Sadly every manufacturer of RF or Matter devices is catering to Amazon/Apple/Google and occasionally Samsung, we are getting left out of some great stuff.  My automatic blinds are a game changer to my sleep and home A/C usage in the summer.

All of my matter devices work great with the Alexa and Apple Home Kit ecosystems right out of the box.  I don't understand why I can poll them with the UDI/eISY but just get "unavailable". I don't want Amazon or Apple knowing when I am home opening and closing blinds, bad enough half of my house uses Alexa for music, but I think I am going to change that.  Half is already wired, so I am thinking of going wireless, non amazon and using my REQUEST Server more or building my own winmedia server since the Request can no longer stream.  I am getting tired of walking in my kitchen to ads about faucets, and my watch telling me I slept in late again.  Just for fun I split my matter devices between the Apple and Alexa systems to see how it affects ads and stuff that shows up at my house.

 

I hear you!

Even though most of the spy style softwares can be disabled the latest trick eveyone is using is

"For our security, we have forced another update on your working system so we can install new features and new methods of disabling everything. Of course, we have turned everything back on, so you can experience the same amount of time we have spent, attempting to force these spy features on you one more time.

We know you are disabling these items and we are watching you"

[CoolToys]

Got a new one tonight.  House was in "away" mode, and I got home.  The program didn't run so I looked in the AC and the "Away to Home" program was yellow.  In the summary, there are two that say "not loaded, out of memory".    The interesting thing is I do not have a program called "Bedtime_2". I built it as a test, and then deleted it after @larryllix cleared up the linear thinking issue.  I rebooted at the bedtime_2 went away but bedtime still says "out of memory"? 

Edited November 15 by CoolToys
update to error

[oberkc]

I would look for two things.  First, did you save all your program changes?  (go back and save changes.). Second, do any of your programs have references to obsolete devices or programs?  (Open the programs in question and see if there is some type of error embedded in any of the clauses.)

Edited November 15 by oberkc

[CoolToys]

@oberkc, Well there is a thinker.  I did a find/replace all on a variable a while ago  I had two that did the same thing, so I changed them all to one, and then deleted the variable from the table.  I also noticed that "Bedtime_2" test program was back.  

Did not know that when you delete a program you must "save" or it is still there just not visible, it doesn't have a "red arrow" to remind me to delete, I thought "delete" was enough.

Did not know that a broken variable can overload the system.  Looking back at logs, there was an extended power failure while I was gone so my guess is the eisy rebooted and brought back my unsaved changes.  

Thank you 

[larryllix]

7 hours ago, CoolToys said:

@oberkc, Well there is a thinker.  I did a find/replace all on a variable a while ago  I had two that did the same thing, so I changed them all to one, and then deleted the variable from the table.  I also noticed that "Bedtime_2" test program was back.  

Did not know that when you delete a program you must "save" or it is still there just not visible, it doesn't have a "red arrow" to remind me to delete, I thought "delete" was enough.

Did not know that a broken variable can overload the system.  Looking back at logs, there was an extended power failure while I was gone so my guess is the eisy rebooted and brought back my unsaved changes.  

Thank you 

You are only editing the IoX copy of the programs so the changes need to be installed into the ISY machine.

Now I am beginning to wonder if the method of updating variable names and values has had the same logic applied to them, where you now need to save them also. That wasn't the case a few versions ago but you may not find them in the pulldown menus of the program editing page.