Hi @chris
I have a couple of Fibaro FGS223 Switches which usually have fast response times. But sporadically there is a very noticeable delay between sending a command to the switch and the physical device responding.
zwave debug shows significant delay between command TX and RX ack
I’m observing this behavior since months with various openhab versions and could not find any solution so far.
currently I’m running openhab version 2.4.0.M4 - but also with your new z-wave binding no improvement
This doesn’t look like a software issue - it just looks like the device is responding slowly. The binding sends the command, and then the device is simply not responding for a few seconds.
This might be because the controller is having to resend the messages which can happen if communications through the mesh is not reliable.
I would take a look at the other devices in the area to see if there is any issue with the links.
Hi Chris,
thanks, I have checked the other devices and have similar issues at least with node 8,9 and 11 (all FGS222 switches)
according to the neighbor list, they are not directly connected to the zwave controller but next hop to a fibaro FGRM222 rollershutter (node 4).
is there any chance to nail down the path or root cause for the intermittent delay for those devices?
How can I establish direct links between controller and nodes? should I try to move the Raspberry Pi or install a dedicated z-wave repeater?
attached the habmin z-wave network viewer
You can’t. The controller will decide the best route through the network - if the link is good directly between devices, then it will use that, but if it’s unreliable, then it will use an intermediate node as a router.
Every mains node is a “dedicated repeater”, and adding a mains socket outlet, or even a repeater (which is basically the same thing, but without the mains output!) in the “right” place should help.
In general, your network looks ok - at least for nodes 8 and 9 - both of these have 3 links toward the controller (eg node 8 links to nodes 3, 4 and 6 which all have links to the controller). Node 11 is a bit more “out there” as it only has a single link to node 4, so could be more flaky.
The network diagram is useful, but of course it doesn’t provide a physical view in relation to the your home so it’s not so easy to provide suggestions on what you can really do. I would suggest trying a mains power outlet in a strategic place to see if it helps.
Hi Chris,
thanks, after installing another FGS223 in the middle of the house (node #17) the situation improved a lot. I would not say it’s solved 100%, but the delay is happening far less
I will try to add two more zwave devices in the next weeks and check if this solves those delays completely
thanks, Oliver
Hi,
I’m new to openhab. I migrated from FHEM.
I have the same problem with my enOcean Wall-Switches and with my Z-Wave dimmers and Plugs…
The delay is about 8 - 10 sec which is far from acceptable.
Position of the Raspberry did not change since the migration…
What do you use to create this “Sniffing” in the Z-Wave network?
Thanks
Robert
Is the delay is right after updating a rule or item file? If yes, then wait some min. to give OH the time to fully reorganize itself.
On my system (win10 with an very old ATOM CPU) the Z-wave commands will send immediatly.
Only dimmers are a problem and I do not use the combination z-wave switch and z-wave dimmer. OH is to slow to send the command within some msec. Direct communication between switch and dimmer will help, but this is not what I want. I want OH to be in the middle every time. Not a big problem for me that only this is not working well.
Hello,
after a reboot or when I restart the service, it takes up to 5 minutes until the BT, enOcean and Z-Wave ist up and running. HUE is after a few secunds available.
When enOcean (enoceanPI module) ist up, it runns without delay. Z-Wave needs additional 20 Minutes to have a good response time.
After a few days, Z-Wave starts to increase the delay again…
I’d like to sniff in the Z-Wave network to find out where the packages are delayed…
Do you know a pice of software which can do that?
Best regards,
Robert
FWIW, at startup the Z-wave binding does a full network scan/repair. It completely floods the Z-wave network until it finishes, and bogs down OpenHAB as well. Its just one of those things you get use to when you restart the system. (Even my slave switches which work directly to their primary switch get laggy during that time, and those don’t even route through OpenHAB – the flood of traffic on the network just slows everything down.)
With about 40 Z-wave devices, it does take about 20-30 minutes for things to settle back down after a restart. So, although I can’t speak to if it should be like that, the behavior you’re seeing is “normal” for OpenHAB. (I’m not sure what the difference is, but both Vera and Homeseer didn’t impact the network that badly… I don’t know if they’re scanning differently, or just being less aggressive about it.)
My system works very much faster as dotorg/KillRob mentioned. Yes, after a restart it takes up to some minutes to be well prepared. But then all went within some sec.
I have a manually reboot test to know if all works well.
I can see in the log that the command is send out and the action is done immediatly.
For me z-wave is very relyable and fast. No difference to my modbus bindind which i use for cable conencted devices.
That depends entirely on how many Z-wave devices you have. The binding walks each one in turn, and has it optimize its routes. How long each takes depends on how many devices you have. So its an exponential slowdown, not linear. Once the optimization process is done, there’s no difference, but the optimization will dramatically impact the network while its running.
For all of you that are interested in my next steps:
Because Z-Wave was not so slow before I migrated to openHAB and because all other things with openHAB are at least as fast as before, I’d like to dug deeper 
I used this Software: http://docs.incontrolha.com/home/remove-a-failed-device
I removed every Node which openHAB did not know as a “THING”.
After that I rebooted my Raspberry. It took about 1 Minute for the Z-Wave Stick to be online and imediatly I was able to contoll my Z-Wave-Nodes !!!
I hope it stayes as fast as it is right now.
Robert
