Enough already

[sjauquet 16]

go to the link in my post above and see the post of 2013-03-30, 17:42 of this thread

[alex88 71]

I have to post that unfortunately after installing a huge antenna yesterday I am still getting dead nodes.

I have managed to debug a lot of the issues (eg nodes that were not physically connected anymore but still appeared in the hc2 interface, faulty nodes that I replaced with new one) and get the count down, but as of this post I am still getting frequent emails with dead nodes.

I am running v3.591.

[Guest shapa]

Just to update - dead nodes issue is 100% Fibaro's fault.

Do not have any single dead node with Indigo6 from scratch (~3 weeks), all devices in a single network (>100 pcs)

The issue is purely related to outstandingly bad z-wave engine made by Fibaro (both v3 and v4)

[blawre0000 1]

Hi

Just to ad my 2 peneth my HC2 has also been up for 21 days sinse I spread out the wake times on my Danfoss TRVs, with no dead nodes. Cant explain it! but it is so.

V 3.591

Brian

[richo 4]

shapa,

Don't have dead nodes problems on Fibaro since 8 weeks !!!

1. Removed all Danfoss valves - dead nodes gone

2. Added better antenna to HC2 (dipol)

3. Included back Danfoss valves (3 weeks ago) - still no dead nodes so far

Please login or register to see this image.

/emoticons/default_icon_smile.gif" alt=":-)" />

Hope it's going to be stable for some time.

Suggest doing the same to the others...see what happens.

Based on my experience dead nodes appear when there is to much "noisy" communication in z-wave network. Reasons are different, some of them:

- to many devices waking at the same time

- to many actions send to devices or reports from devices at the same time

- devices that have problems with communication (panic mode in Danfoss)

- badly written scenes that generate to much network traffic

- to many changes in the network and reconfigurations

- problems with z-wave database on HC2

Fibaro z-wave management add to the problem but is not the only source.

PS

In my opinion Danfoss valves have to have DIRECT communication with HC to work properly.

PPS

Have over 100 devices (12 Danfoss valves) plus 20 VDevices and 35 scenes.

Please login or register to see this attachment.

[alex88 71]

I have to post that unfortunately after installing a huge antenna yesterday I am still getting dead nodes.

I have managed to debug a lot of the issues (eg nodes that were not physically connected anymore but still appeared in the hc2 interface, faulty nodes that I replaced with new one) and get the count down, but as of this post I am still getting frequent emails with dead nodes.

I am running v3.591.

Well I have been developing lighting scripts today and I KEEP getting dead nodes in my kitchen. I have two fibaro modules sitting right next to each other. One keeps going dead, the other seems to hold on.. but very occasionally goes dead as well.

I swapped the dimmer that kept going dead for a new one, but have the same problem. It seems to happen when I turn the lights on/off from within a scene. I have added sleeps in my on/off commands in case that was the problem but still the same problem.

[Guest shapa]

richo - "Added better antenna to HC2 (dipol) "

Wonderful. Therefore we can assume that Fibaro's antenna is rubbish and all HC2 units sold (as well as HCL) are broken?

So naive

Please login or register to see this image.

/emoticons/default_smile.png" alt=":)" srcset="https://forum.fibaro.com/uploads/emoticons/[email protected] 2x" width="20" height="20" />

With Indigo6 I don't have any extrenal antenna at all. Small USB key with built-in antenna. All z-wave devices, including Danfoss valves, works just fine (12 valves BTW)

With HC2 / HCL the only way to stop devices to die is to switch off this functionality completely.

It's not about congestion at all, as My fibaro network was split by 3 parts and still a lot of dead nodes.

Now I've got all my devices in a single network managed by Indigo - no single device dead.

[alex88 71]

With HC2 / HCL the only way to stop devices to die is to switch off this functionality completely.

What is the downside of switching the functionality off other than presumably you won’t know if the node really goes dead?

[j.nowacki 5]

amilanov, if the nodes are really dead but are not marked as such the controller tries to contact it every time when it is meant to. That can cause much traffic, because once controller can not reach destination node it tries to find alternative paths in Z-wave mesh network, sending a lot of frames to many devices. And increased traffic can lead to other issues with Z-wave network. When device is marked as dead, controller do not perform attempts to contact it unless it is asked to do so.

[alex88 71]

Ah, ok. Good explanation, so I won't be changing the settings anytime soon as I already have too much network traffic.

[Guest shapa]

amilanov - in fact, they used to work normally. For a while. The best result I've seen - 2 weeks without major problems. Then HC2 hangs

Please login or register to see this image.

/emoticons/default_smile.png" alt=":)" srcset="https://forum.fibaro.com/uploads/emoticons/[email protected] 2x" width="20" height="20" />

Indeed, dead nodes detection algorithm is broken. The issue is in the z-wave stack made by Fibaro, especially in v3 (single thread).

When fibaro is trying to claim that this issue happens due to heavy network traffic - it is a false-claim.

Again, no any issues with much bigger network managed by different controller brand. Obviously, the same house / environment.

[richo 4]

shapa,

Does Indigo and your z-wave controller have the option to mark dead devices?

[Nickwren 4]

They can add me to the list of people leaving. Had enough of waiting and promises of features. Even got sucked into there way of testing software - let the customer do it!! Thats not a professional way to approach development. I manage development of variouse controllers for car body systems and we don't send code out for the customer to have a play with it and find all the issues.

[fergalmccarthy 5]

shapa,

Does Indigo and your z-wave controller have the option to mark dead devices?

richo, there is no need for this option because there are no dead nodes in Indigo, thats a Fibaro thing. I've 100+ zwave devices in Indigo, majority battery powered, I'll let you take a guess at what percentage are "dead" in Indigo.

By the way, the same 100+ devices had about 14 - 20% dead devices in Fibaro, both "stable" and beta versions. The need for more than one fibaro controller for a zwave network in a house is comical to me and is just a sales tactic. I have one 60 euro Z stick in a mac and that is all I need. The whole purpose of a mesh network is for powered nodes to relay the signal so with some proper planning there should never be a need for another controller

[Dino 0]

Yep yep still dead nodes in de revamped z-wave engine of 4.018 beta. But hey my misteroius Fibaro Ghost has gone; switchen "on" of devices with no action. For now than and lest hope they will not get back.

In other words beta 4.018 for me is more stable than 3.591. And I think 4.018 is close to be called a beta. But the development proces is one of amateurism.

I just need 0-10v support on the RGBW module in Indigo.

[Guest Kuuno]

Yep yep still dead nodes in de revamped z-wave engine of 4.018 beta. But hey my misteroius Fibaro Ghost has gone; switchen "on" of devices with no action. For now than and lest hope they will not get back.

In other words beta 4.018 for me is more stable than 3.591. And I think 4.018 is close to be called a beta. But the development proces is one of amateurism.

I just need 0-10v support on the RGBW module in Indigo.

we have this

Please login or register to see this image.

/emoticons/default_smile.png" alt=":)" srcset="https://forum.fibaro.com/uploads/emoticons/[email protected] 2x" width="20" height="20" /> pm me