Has anyone done a successful backup of Aeon Labs Gen5 Z-Stick?

No - this won’t work as you will have two nodes with the same node ID. This can’t be a good thing.

It’s like having two computers with the same IP address - when you send a message, which one will answer? Both probably.

Of course they know which node sent a message - otherwise they would never know where to respond to.

In any case, a controller is not a slave node.

Both sticks MIGHT receive the same message - if they are right next to each other then there’s a reasonable chance this is true. However, both sticks will then respond which will cause confusion. Even worse is if one stick only receives the message (due to RF, it’s not guaranteed that both sticks will receive the same packets). Then you really have a mess…

Given both sticks would think they are the primary controller on the network, I think there would be a major mess.

I still have a hard time to follow your thoughts. So you assume to have 2 OH installations running at the same time on the same location? If both are firing the same rules at nearly the same time, what do you expect? How are you going to organize the “hot redundancy”? The 2nd stick and the 2nd installation should only take over if the first fails, correct?

  1. How do you get the information that installation1 failed and
  2. how do you trigger installation1 to shut down in this circumstances and installation 2 to activate?

I also think - like Chris - that this will not work.

Before we continue let’s discuss simple setup:
Imagine you have two cloned USB Z-wave sticks inserted in same PC and they shown as COM1 and COM2. Openhab communicates with one of them, for example on COM1 - let’s call this “active” USB stick. Another USB stick on COM2 - let’s call it “passive”, just sits there and does nothing, as Openhab does not communicate with it.
If something happens with active USB stick - for example it stopped reacting on COM commands, Openhab just failovers to passive stick, by changing the COM port to COM2. I know, that this is not realized yet, but sounds rather simple, isn’t it?
So now the questions:

  • will passive USB stick disturb the z-wave network so much, that it can not be kept in such mode? The COM communication is off for it, so it should not issue any commands.

I do not know if it would disturb the communication. So far I only had one USB stick “on power” (in the plug) while the other one was just lying around. This does not do any harm (although these devices have a battery).
As the COM port for Z-Wave binding is defined in the openhab.cfg file the “hot plug function” would mean that openhab is modifying on its own the openhab.cfg file to another Com port if it detects that the controller fails (whatever the condition should be for this). Quite a connection of a lot of weird steps IMHO.

Normally the stick will work for years without any trouble. If however it should fail one day, well, then you will have to unplug the old one and plug in the backup stick. This is for me the absolutely accectable solution while towards my mind the theoretical discussion of a whatever “hot plug” function is not necessary.

I don’t want to discuss this now - this is just one possible implementation from many possible solutions.

For me it is not - what if it happens when I’m on vacation? In my case Z-wave controls pretty everything without any other control capability except cutting the protection circuit breakers. I don’t have any mechanical switch in my home.

Good idea…

Yes. Just because the COM port is not used, doesn’t mean the node isn’t doing anything - the device is autonomous.

A ZWave node that is powered will participate in the network. It will route packets for other nodes, it will respond to any request that is sent to it. As a controller, devices may send various updates, and if there is another controller in the network that is sending commands, then this device that has the same node ID will receive all these responses, and it will reply to them all. This will cause a mess on the network.

Ok. So the next question would be - is it possible to disable temporary any of these functionalities on Aeon Z-Wave USB stick by COM port command? E.g for example by putting it into some mode or disabling z-wave transmitter? At least I have seen transmission power register in Aeon Z-wave utility - not sure if it can switch transmitter off completely.

No. You would need to power the stick down.

Controllers do not respond to command classes - ie they don’t implement the command classes, and you can’t send them any Z-Wave command class via the serial port. Maybe Aeon have a secret command to do this, but given that this interface isn’t documented, I can’t really comment on this.

However, even if you could reduce the power, I don’t think it would solve the problem unless the second controller is quite a long way from every other device and given that if you were to move the two sticks to different places, routing wouldn’t work, I don’t think you can do this.

You cannot control the light at home from your holiday location. So what?
As soon as you come back home you plug the backup stick in and everything runs smoothly again.

In this case you perhaps should reflect your architecture. In my case I intentionally implemented a “normal” installation being capable to control everything also with normal switches etc. So even if my server or the Z-Wave controller fails everything works (except of course some automated rules, but I still can control the light, the heating and the rollershutters).

Well, I have a bit more than just light. For example z-wave controlled heating - if USB stick fails, while it’s on and I’m on vacation I risk to receive quite a good Bill for heating at least. Also this can happen while I’m not at home, but someone else - kids, grandparents. They are not able to replace USB stick, but still need to control light, shutters etc

1.) Sorry, but I do not understand your point for heating: My Z-Wave thermostats just keep the defined temperature. I only need Z-Wave communication to CHANGE the set temperature (and I could even change it by wall switch). So your only problem is that you cannot change the set temperature by smart phone from your holiday location. So what?

2.) Anyhow, that’s why I have chosen an installation with Z-Wave that still is compatible with kids, grandparents etc…and I informed myself about all this before doing the installation. Why do you have no wall switches? Does not sound suitable for normal life operation to me…

3.) The kind of hot plug functionality you are aksing for is even not implemented in many industrial server landscapes and as this is my business for more than 20 years I know what I am talking about. In most cases the admin has to switch manually to the Backup Solution and this is acceptable.

4.) Even if the Hot-Plug-Support would be implemented in the future, what would be the IF-condition that it should be activated? How can the software determine reliably that the stick has “failed”?

The discussion here about “Hot Plug” is IMHO purely theoretical without any relevance for practical use. The availiablity of a backup feature für the Aeonlabs Stick is already a quantum leap compared to all toher Z-Wave solutions on the market and only available since 2016 (while the stick was on the market many months ago). With all other controllers you will have to manually exclude and reinclude ALL NODES of your network which (with 100 nodes) will take you at least a weekend.
So it is nice to demand for “hot plug” for the future but actually this is state of the art.

Just my 5 cents…

  1. I don’t have Z-wave Thermostats on the walls. They are expensive and unflexible. Instead I have Z-wave Motion Sensors, which provide temperature/humidity readings as well. On floor heating valves I have standard Z-wave relays. Temperature settings are defined by control panels (Android) or Web or phone and by schedule. Finally heating algorithm executed by main Controller. Currently I have simple hysteresis with night setback, in future I think to add Nest-like learning algorithm.
  2. During normal operation my installation is quite compatible with kids/grands too. All control is done by intuitive touchpanels with individual GUI design. Also some controls are duplicated on different panels for flexibility. No wall switches - I have a new apartments with all electricity. Imagine in my living/dining room I have: 7 Blinds, 1 dimmed main light, Kitchen LED Spots, Kitchen LED Table Light, RGB LED Ambient light and few other lights in plan. How many wall switches I need, and how I allocate functions on those? 1,2,3 pushes? What about RGB and Dimmers? I did had axperience with mess of 6 switches on the wall - didn’t like it.
  3. Really? Maybe I have seen different industrial server environment, than you. In mine RAIDs are standard. Dual hot-swap power supplies are standard. Fans are redundant. Ethernet ports are dual. Yes, sometimes manual switch is required, when you switch keyboard and Display to backup server using KVM. This is the case when servers are redundant, but keyboard and display not - just to save some space in cabinet.
  4. This is relatively easy task - for example pings are already implemented.

For me this is a practical solution, which I’m going to implement anyway. This way or the other. If this doesn’t work, other solution would be to use switchable USB hubs - they are available. Concerning backup quantum leap feature is not true - as far as I know, Home Centers like Fibaro or Vera can do backups including Z-wave configurations since years. Also as far as I remember Razberry can do backups as well.

PS If you are interested I propose to continue discussion in my topic: OpenHAB Server Redundancy - any Ideas?

@chris, I still performed such experiment. And you know what? Nothing happened! E.g. I cloned USB stick and inserted it in my laptop, which was running IMA Tool. As the result second Stick was powered up and showed all discovered nodes. At the same moment I monitored Z-wave logs in OH from another stick and seen nothing strange there. The system was behaving exactly as before with no latency. Only when I started pinging process from second stick, Z-wave logs were showing that discovery messages are coming. But that’s all. IMA tool also showed, that my temperature sensor reports are received by second stick too.
I don’t know if there will by more side effects, but now it seems to be working quite nicely.

Great - just don’t be surprised if there are random problems experienced since there WILL be acknowledgements being sent from both sticks. If you’re lucky, it won’t cause a problem, but if there are events like routing updates, neighbour updates etc, that generate a lot of traffic, then who knows what will happen. It might work ok, or you might end up with confused routing if one stick or the other ends up responding first.

Good luck :wink:

Is it possible to see those acknowledges in Z-wave log? I don’t see them now there and would like to check.
Concerning routing and neighbour updates - I have disabled those since all my nodes are fixed and I prefer to do manual updates only. I found this is more reliable way of doing such thing.

No - these are low level protocol messages - they aren’t communicated to the computer. There are a lot of things happening that you won’t see at all in the log since the stick is autonomous in many areas. Devices will periodically communicate with the controller, and with two controllers using the same address, they might respond differently.

Maybe, but the system uses explorer frames by default…

All I can say is it may work, or it may not. If you see strange things happening, then don’t be surprised as fundimentally, this simply doesn’t sound good from a protocol perspective. It’s the same as having two computers using the same IP address - it might work, or it might not…

How about an alternative:

  • Main PC as normal, i.e. with one z-wave stick, plugged into a wemo (or other wifi switch) that is ordinarily ON
  • Backup PC (or raspberry pi) with cloned openHAB installation and cloned z-wave stick, attached to a wemo that is ordinarily OFF.
  • If main PC reports zwave stick has failed, main PC is shutdown and backup PC is started

Not the most efficient setup, but gives you redundancy!

Dan

There are many alternatives to create redundancy based on current findings - for example many popular switchable USB hubs can power down and up individual USB ports by Linux command, and all this works in Rpi. All I wanted to know if powering off really required or not.

Finally, the most important thing, which we have found here, is that Z-wave Controller can be made redundant at least using AEON sticks. This makes Z-wave suitable for more high-critical home automation systems, which were always prerogative of more high-price distributed systems.

This is exactly the point: What happens if a piece of software or hardware fails? My installation is more tolerant and reliable in this case.

I totally agree that the optical appearance is not optimal. However, this gives me already a lot of the redundancy in all situations that your installation is lacking.

Ok? Is there any command which can be used in an OH-rule to ping the controller and determine if it is dead?

For the commercial solutions this is true. However, for open solutions like OH the options were rare. The Z-Way-Stick offered this option first, but the backup/restore-software only worked on some Linux distributions. As I am running a windows server, the Aeonlabs Gen5 Stick is (as far as I know) the only available option.

I absolutely agree with you in these points. And I also like your ideas for the future developments. All I wanted to point out is that at the moment (and to my knowledge) a solution as you describe is not yet available out of the box. And current OH bindings and architecture do not support such scenarios.
BUT: It would be definetely great if such a scenario could be realized in the future.

Yes, we are moving more and more off topic here. So the conlcusion is: Backup (and restore) of Aoen Labs Gen5 Stick works absolutely fine. A Hot Plug Z-Wave-Controller solution would be the “next big thing”, but not in this thread.