Sensor unplugged, but SDJ says it was successfully polled?

[mlevin77]

1st floor Gym Gym motion/light sensor - Parent with 4 children Last Change Time yesterday at 8:33 AM - Period = 1 days 0 hours 10 mins

Exceded inactivity period but successfully polled

Says it was successfully polled, but I unplugged the sensor yesterday (> 24 hours ago), to make sure I get an alert if it goes down. Interestingly, the main page of HS3 was showing status RED until I tried to manually test connectivity (which resulted, correctly, in Failed to reach node 15 (1st floor Gym Gym motion/light sensor)) and now it's showing as "Unknown" on the main page, and now the failure report lists it correctly as being in error. How come it didn't find it in alert until I manually tried to test connectivity?

[SteveMSJ]

1st floor Gym Gym motion/light sensor - Parent with 4 children Last Change Time yesterday at 8:33 AM - Period = 1 days 0 hours 10 mins

Exceded inactivity period but successfully polled

Says it was successfully polled, but I unplugged the sensor yesterday (> 24 hours ago), to make sure I get an alert if it goes down. Interestingly, the main page of HS3 was showing status RED until I tried to manually test connectivity (which resulted, correctly, in Failed to reach node 15 (1st floor Gym Gym motion/light sensor)) and now it's showing as "Unknown" on the main page, and now the failure report lists it correctly as being in error. How come it didn't find it in alert until I manually tried to test connectivity?

Hi mlevin77,

That's interesting you appear to have picked up a bug that didn't exist when I last tested!

I've now done some more testing and it appears that there is a bug in the Z-Wave plug-in whereby, if you poll the parent device it returns a valid result even if the device is unplugged. I'm sure that didn't use to be the case, but I can't say what Z-Wave pi version it might have been introduced. HS are bogged down with HS4 troubleshooting so the easiest thing is for me to modify my plug-in to poll one or more of the child devices rather than the parent. I'll let you know here when I have a beta version updated for testing.

This only effects the 'General Devices' section of the plug-in. Any polling carried out by the 'Battery Devices', which most people use, is carried out on the battery child.

Well spotted

Steve

[mlevin77]

Remarkable :-) I was sure I was doing something wrong. Thank you for your help on this! I'll be watching for word of the new version.

[SteveMSJ]

Remarkable :-) I was sure I was doing something wrong. Thank you for your help on this! I'll be watching for word of the new version.

Unfortunately it seems this is a more serious bug in the Z-Wave pi than I at first thought.
For testing I have been using a TKB TZ69E plugin switch, as it is easy to unplug.
It has a Parent and 3 child devices:


If I physically unplug it from the wall and then poll it from the Z-Wave pi (click on the Poll icon) it only throws an error for the Switch child, the Parent and the 2 Energy Sensors return a result without error. The results are all a value of zero but that is the correct value for #202.


I'm sure this is a new bug in the Z-Wave pi. I will report it to HS but I don't expect a rapid response or resolution.

Can you carry out this check on your sensor and see what result you get. That is, in HS3 display the group of devices as above and click on the Poll icon/button, second from the right at the top.

The only work around I can think of at the moment is to poll all the devices, Parent and Children, to check that they all respond without error, but I don't really want to do that as it increases the amount of polling significantly.

Steve

[mlevin77]

they all update right away and here's what I see (the device I'm referring to is in lines 1-5):

[SteveMSJ]

they all update right away and here's what I see (the device I'm referring to is in lines 1-5):

So in your case none of the child devices returns an error when polled even though the physical device is powered down?
Is that all the devices representing the physical device in the first 5 lines?
The 6th line starting '3 @' does show 'Result not set by Plugin' although not actually an error. Is this a different device?

That makes checking whether a powered device is alive, that doesn't show regular activity, somewhat difficult

Steve

[mlevin77]

ok hold on, sorry, you wanted the device unplugged, I forgot about that. (and "3" is the same device)
Here it is when unplugged:


lines 1-5 are the device I unplugged. Looks like only the children on 1 and 2 are showing the fact that it's off the network!

[SteveMSJ]

Unfortunately it seems this is a more serious bug in the Z-Wave pi than I at first thought.

The only work around I can think of at the moment is to poll all the devices, Parent and Children, to check that they all respond without error, but I don't really want to do that as it increases the amount of polling significantly.

I set up a fresh install of HS3 with a new Z-Wave network so I could do some more comprehensive testing of the Z-Wave pi polling various devices in powered and unpowered states.

Firstly my findings show that this doesn't impact on the Battery Devices section of SDJ-Health, which is what most people use. For Battery devices that are 'listening', such as locks, if you choose to test by polling the pi polls the battery child. My testing shows that the Z-Wave pi does correctly return 'unsuccessful' for the battery child on the battery devices I tested with the batteries removed.

However, for powered devices the Z-Wave pi incorrectly returns 'successful' for an unplugged device for the Parent and many of the Child devices. It does correctly return 'unsuccessful' for some child devices such as switches. I haven't yet been able to determine the exact type of devices for which it responds correctly.

I will report this to Homeseer but I don't expect it will get looked at any time soon so I am working at a work round.

What I have found is, for an unplugged powered device, if you poll the parent you get an instant incorrect 'successful' response but for the child devices the incorrect 'successful' response is not returned until the Z-Wave 'Send Timeout' period is reached. This is a default of 4 seconds but can be edited by the user. So the Z-Wave pi waits for the Time Out period and then returns 'successful' even though it hasn't got a valid response for most child devices. Not good.

What I am looking at doing is timing the polls and if they reach the Timeout, I will class that as a fail even though the Z-Wave PluginFunction result is 'success'. As the user can change the default I will read the Z-Wave 'Send Timeout' from their Z-Wave.ini file. This should generally work ok.

When I have a beta for testing I will update this thread.

Steve

[SteveMSJ]

I set up a fresh install of HS3 with a new Z-Wave network so I could do some more comprehensive testing of the Z-Wave pi polling various devices in powered and unpowered states.

Firstly my findings show that this doesn't impact on the Battery Devices section of SDJ-Health, which is what most people use. For Battery devices that are 'listening', such as locks, if you choose to test by polling the pi polls the battery child. My testing shows that the Z-Wave pi does correctly return 'unsuccessful' for the battery child on the battery devices I tested with the batteries removed.

However, for powered devices the Z-Wave pi incorrectly returns 'successful' for an unplugged device for the Parent and many of the Child devices. It does correctly return 'unsuccessful' for some child devices such as switches. I haven't yet been able to determine the exact type of devices for which it responds correctly.

I will report this to Homeseer but I don't expect it will get looked at any time soon so I am working at a work round.

What I have found is, for an unplugged powered device, if you poll the parent you get an instant incorrect 'successful' response but for the child devices the incorrect 'successful' response is not returned until the Z-Wave 'Send Timeout' period is reached. This is a default of 4 seconds but can be edited by the user. So the Z-Wave pi waits for the Time Out period and then returns 'successful' even though it hasn't got a valid response for most child devices. Not good.

What I am looking at doing is timing the polls and if they reach the Timeout, I will class that as a fail even though the Z-Wave PluginFunction result is 'success'. As the user can change the default I will read the Z-Wave 'Send Timeout' from their Z-Wave.ini file. This should generally work ok.

When I have a beta for testing I will update this thread.

Steve

mlevin77

I have posted an updated version 3.0.8.8 as a beta which has the work round for the Z-Wave poll bug. Please could you give it a try, testing as you did previously, and see if it works for you this time?

Thanks,
Steve

[mlevin77]

mlevin77

I have posted an updated version 3.0.8.8 as a beta which has the work round for the Z-Wave poll bug. Please could you give it a try, testing as you did previously, and see if it works for you this time?

Thanks,
Steve

sorry, where do I get the beta? I only know how to install things when they appear on the plug-in page as new versions.

[SteveMSJ]

sorry, where do I get the beta? I only know how to install things when they appear on the plug-in page as new versions.

In HS3 if you go to Add Plug-ins and at the bottom of the page there is a section for Beta versions.
Steve

[mlevin77]

Thanks. I just ran it, a few things I don't understand. First, it just alerted me to this:

Garage Garage HomeSeer Tech Notification Sensor Node 8 Root - Parent with 4 children Last Change Time 11/17/2020 at 10:18 AM - Period = 42 days 4 hours 0 mins

Exceded inactivity period and failed poll

which I confirm is true on the devices page:

I guess it's working to let me know, what I've never noticed, that the child nodes claim to have a last update on 11/17, but this makes no sense as the motion detector there is working perfectly and was triggered plenty in the last couple of days. What does this mean?

[mlevin77]

Also:

1st floor Gym Gym motion/light sensor - Parent with 4 children Last Change Time today at 2:17 PM - Period = 0 mins 48 secs

Passed activity scan but Invalid State

this can't be right because that sensor is unplugged (I unplugged it several days ago when we first started to debug this). here's what the devices page shows:

the color control is from me trying to change the color just to see what will happen (I clicked the color and of course nothing happened as the thing is unplugged, but I don't know why it updated the access time?), and the report says it passed the activity scan, probably from this bug in HS reporting an updated time when nothing successfully happened? The children, having been inactive since 12/26, didn't trigger it as a fault? I assume at least the change of temp or light levels, which happens many times per day, should trigger a problem if the date is > 24 hours?

[SteveMSJ]

Thanks. I just ran it, a few things I don't understand. First, it just alerted me to this:

Garage Garage HomeSeer Tech Notification Sensor Node 8 Root - Parent with 4 children Last Change Time 11/17/2020 at 10:18 AM - Period = 42 days 4 hours 0 mins

Exceded inactivity period and failed poll

which I confirm is true on the devices page:

I guess it's working to let me know, what I've never noticed, that the child nodes claim to have a last update on 11/17, but this makes no sense as the motion detector there is working perfectly and was triggered plenty in the last couple of days. What does this mean?

What was the time of the scan on this report?

Steve

[SteveMSJ]

Also:

1st floor Gym Gym motion/light sensor - Parent with 4 children Last Change Time today at 2:17 PM - Period = 0 mins 48 secs

Passed activity scan but Invalid State

this can't be right because that sensor is unplugged (I unplugged it several days ago when we first started to debug this). here's what the devices page shows:

the color control is from me trying to change the color just to see what will happen (I clicked the color and of course nothing happened as the thing is unplugged, but I don't know why it updated the access time?), and the report says it passed the activity scan, probably from this bug in HS reporting an updated time when nothing successfully happened? The children, having been inactive since 12/26, didn't trigger it as a fault? I assume at least the change of temp or light levels, which happens many times per day, should trigger a problem if the date is > 24 hours?

Are you saying that this device is unplugged yet HS3 updates the Last Change Time of the Color control? That sounds like a bug!

As long as one devices representing a physical device has updated it shouldn't trigger an alert.

By the way what exactly is this device? Can you give me the model number?

Steve