Why does the nightly automatic heal make my network worse?

I have a few questions about the automatic heal. I think there is a serious problem with this process:

  1. How can it be the automatic heal makes your network perform worse than before the automatic heal?
  2. Why is it possible to manually repair the network after an automatic heal by just healing the not configured devices? Seems a bit silly to correct an automated process.
  3. Why isn`t there an option to disable the automatic heal?
  4. Do more people have this problem?

Thanks

in what way does it make it worse?

Worse as in devices not being configured after heal, devices previously associated not responding any more, errors like can’t get name. It seems to me a heal should not corrupt a working network.

do you have a lot of devices? did you give the heal process enough time to complete? before the heal, does everything work fine?

What type of devices? Is it always the same ones? Are these devices battery operated devices? The more details you can provide us the better we can help.

  • Garrett

I have 6 x duwi shutters, 4 x duwi switch, 2 x duwi dimmer, 3 x fibaro double switch, 2 x fibaro single switch, 2 x greenwave 6 plug, 1 x everspring AN158-2, 3 x duwi wall controller, 2 x EZmotion 3 in 1. Before heal all works fine, but after heal often a a couple of shutters will fail and the association between a duwi switch and a fibaro switch is lost. The switches which fail are not located at a specific place. In all cases the is another z-wave device within 5 meters. The heal problems don’t concern battery operated devices.

The strange thing is a manual heal of the ‘bad’ devices (one by one) fixes the network!

Thanks

this process (nightly heal) is known to mangle battery devices, I have a vera3 with only 2 devices on it added yesterday and the 2am “heal” killed my 3in1 sensor

In my case most of the time powered devices are corrupted. Battery devices are sometimes corrupted.

@all: does anybody also has problems like this?

I agree on this automatic heal.
You can easily turn it off.
Go to Setup–> Zwave Settings and disable: “Use Vera routing instead of Z-Wave (requires 4.5)”.

I have disabled this option and my network is running much much better.

@ frichter09: Why did you disable the automatic heal? Did it corrupt your network too?

Nope,
It did not corrupt anything.
I have about 20 devices and sometimes - for whatever reason - there is one device that has one or two failed polls during the day.
Vera then starts to do the nightly heal. I just think that this is unnecessary - I like to know what Vera does and when. I do not need a complete heal every night or every other night - thereby using a lot of my locks battery…
Thats why I turned it off…
Flo

[quote=“frichter09, post:11, topic:172220”]Nope,
It did not corrupt anything.
I have about 20 devices and sometimes - for whatever reason - there is one device that has one or two failed polls during the day.
Vera then starts to do the nightly heal. I just think that this is unnecessary - I like to know what Vera does and when. I do not need a complete heal every night or every other night - thereby using a lot of my locks battery…
Thats why I turned it off…
Flo[/quote]

I thought Vera only did heals when it saw a problem not just automatically

Hi aschwalb,
yes, you are correct.
So the “problem” in my case is that sometimes Vera has an issue polling a device.
This does not happen often but sometimes. Vera marks the route with an X and there you go.
A heal will be performed at night.

Thats why I turned it off.

I know I am late replying to this thread but I have a similar issue - most of the devices work fine in my house but I have a shed that is about 20m from the house. Looks like any fibaro devices in the shed do not update their routes as expected - yet the everspring plug I have in the shed works perfectly every time. From my basic understanding it looks almost as if the fibaros are not relaying properly.

This is a case where we would probably like to get a notification from Vera … Not just an automatic FIX It’s terrible to wake up in the morning and find most of your devices flagged in red with a neighbor nodes error on the UI5 webpage.

Ok - I have spent hours on this now and have come to one conclusion: The documentation for vera/zwave routing is terrible!!

I am still testing but disabling vera routing definitely stops the overnight heal from breaking things. I know the heal is supposed to only kick off if it cannot reach devices but it seems the fibaro’s are not that compatible when it comes to zwave relay.

I dont really understand this - so if anyone can shed light on it then please let me know - I have searched the forums and this seems to be the syntax for manual routing is a dot sepparated list eg:

2.4 - where you want to go from vera → node #2 → node #4 → end node (in my case #16)

In my case I have a number of nodes that #16 (the fibaro switch) could relay off but it doesnt. It’s also the only node that is too far away to directly talk to the vera.

There are several nodes in range of node #16:
Everspring plug: #27
Everspring Plug: #4
Fibaro switch: #29
Horstmann boiler controller: #15

logically based on physical layout the route should be:
15.29
or: 29.4
or 29
or 27
or possibly 4 (it’s a bit of a stretch!)

so I tried writing the route as: 15.29,29.4,29,27,4
That didnt work.

Then I tried ‘4’ - That didnt work either.

In fact the only thing that will reliably relay to the fibaro switch is another fibaro - node #29 (manual route simply written as “29”) - it wont realy from the eversprings and it also does not seem to be able to relay from two fibaros - it has to be just one hop away.

Has anyone else done any testing on this? or am I doing something very wrong - probably as I said the documentation is awful at best.

I have a spare aeon labs dimmer that I am going to replace node #16 with to see if its a general routing issue or just a quirk of the fibaro units.

The other thing that is confusing me is that after a while the neighbour nodes for #16 are defined as 1,15, even after defining a manual route of #29 - why is 29 not in the neighbour nodes list? and why is 4 not there - its abut 1m away and by far the closest node!?

I suspect that when left to it’s own devices the routes that the vera initially discovers are not all valid - or the reception is intermittent. SO when it does it’s nightly heal it detects some routes are no longer working, does a heal and then breaks it even more - maybe it’s preferring routes that do not involve relaying? this would kinda make sense, but when it’s doing the heal the house is electrically ‘quiet’ but during the day there are microwaves, ovens, washing machines and so on that may be interfering ans so reducing the effective range - shorter hops would be more reliable.

Confused,

???

Have you looked here:

http://forum.micasaverde.com/index.php/topic,9951.msg67470.html#msg67470

  • Garrett

I did - thanks, but I missed the last post that points to the wiki - http://wiki.micasaverde.com/index.php/ZWave_Debugging#Manual_routing

I missed the AutoRoute variable completely. But looking at it is confusing me even more:

Autoroute for node 29 has one route - 0-5x so I am guessing it’s basically saying its not relaying but it’s currently broken - yet the device is working fine and I can control it. (no manual route defined)

Autoroute for node 16 (the shed light) gives the following: 15.29-30x,23.29-30x,27.29-30x,27.15.29-40x which looks like there are 4 routes - none of which are functional - yet it’s actually working perfectly (I have a manual route of 29 configured)

Also for node 29 - nodes 15 and 27 are closer so if the direct route failed then why didnt it discover the routes using those two nodes?

When I have a bit more time i’ll kick off a full heal see what happens with that manual route in there…

Thanks,

Sam

Just to update… After specifying manual routes I turned on the autoheal again. It’s been about 4 days and the devices in the shed still work fine. I then took out the manual root yesterday and the autoheal broke it again :frowning:

So there’s your choice - specify manual routes for those devices that may be out on a z-wave limb, or disable the autoheal…

That’s the way I have mine configured (thanks to help from frichter09) and everything has been working pretty well. It was a frustrating learning curve for me.