Just had a device die with a completely dead battery, though the logs were showing 100% like most other devices on my system. It was an aeotec dry contact sensor.
Iâm not sure what response youâre expecting? If the device is reporting 100%, then the binding canât do much about that . Or do you mean the device was reporting a low number but the binding was still saying 100%?
HI, hmm wondering what im doing wrong then.
already excluded and re-included it.
but i have strange things in my log, including entries like âNode deadâ or âIs currently marked as failed by the controller!â.
but with the non Security binding its working.
i also noticed that all my battery devices are showing as " node is not comunicating with controller" in habmin. havenât hat that with the ânon securityâ binding.
Hi @chris, do you have any suggestion for me what the problem could be with the door/window sensor?
after some additional âHealâ for those two battery devices they at least showing to be working fine in HABmin, however the door/window is still not reporting any open/close state at all.
i tried again with the binding included in the snapshot release of OH2 and there everything is fine.
i re-included the node and this is what i can see if i open/close it"
This means that the controller thinks the device is no communicating. There was a long discussion on this earlier (somewhere!) in this thread. I donât really know why the controller sets this state on devices that are working - according to the docs, it shouldnât. This version of the binding uses the controller status to decide if the device is dead or not rather than doing something internally - this is why the older version works.
Yes, there was a discussion here on this. But I remember that you planned to change this as the controller status seemed not to be reliable. Do you still plan to do this?
At the moment my âworkaroundâ for main powered nodes (230V, no battery devices) that are marked as âdeadâ is to change an unimportant configuration parameter in Habmin. In this case the binding apparently seems to ignore the dead status, sends the parameter and gets a confirmation. Immediately the system starts the mesh update for the node and it works again. It is silly that these 2 clicks immediately âcureâ each main powered node while the binding will show it for hours or days as âdeadâ.
However, for battery powered devices this does not work. I have a couple of sensative strips door/window sensors which are dead since more than several months (according to the binding and also according to the last time stamp of the xml-File in the Z-Wave folder). However, they report their status several times a day reliably to OH.
So I would like a solution that does not ignore these facts but states a ârealâ status of the devices (as in former times, eg. OH1, OH2 etc.)
Iâm not sure I understand this? What is the âRealâ status? The theory is that the controller has the best information, and therefore it knows the ârealâ status better than anyone else. The binding can try to evaluate the status, but Iâm not sure what you refer to as a ârealâ status? I guess you are asking to ignore the controller? Either way, we have to ignore some âfactsâ?
Can you explain what you mean? (sorry for my lack of understanding with this).
as per my previous post it seems that the controller is now not thinking any more the Node would be dead and as it looks like i receive messages from the node when the contact gets opened/closed.
Following another topic, Iâve installed the latest zwave binding. In short, after a while, my zwave setup stops working, all other things (knx, weather, pushoverâŠ) worked.
No idea why zwave goes dead. Iâm running OH2 on Ubuntu17.10 as VM under proxmox. And Iâm using 2(!) GEN5 USB sticks. This worked for about 1 year without problems. Itâs possible that itâs a hardware/KV issue. But after 4 complete reinstalls (OS and OH), reset of 30 zwave nodes, I donât have any clue more how to proceed.
I canât see this from the logs though - the last logs you posted say itâs failed.
What I see in the logs is that a command class isnât found - this might be because the initialisation hasnât completed due to the FAILED step timing out. Without a full log itâs hard to say why this class is not found.
You could try a complete init of this node to see if that fixes the issue. If not send me the full logs.
Again, without logs Iâm just guessing, but possibly these commands are sent to the root node and the command classes get added even if the initialisation didnât detect them. It doesnât necessarily mean the initialisation completed.
My impression is that the current status handling of dead nodes in the dev binding is not reliable while old versions (OH1 or OH2 non dev versions) appeared to be more reliable. I understood that you implemented your âownâ status handling on the binding side in these old versions while you rely on the controller information in the dev binding.
So I would suggest to reimplement your âown status handlingâ if possible as from my experience it used to be superior to the current one.
@chris thanks for looking into this.
i attached the log from today. i tried several times to reinitialize the node (node 33) but the
COMMAND_CLASS_MULTI_CHANNEL is still showing as not found.
The log doesnât really have much in it. Itâs just an event log rather than an actual log⊠What Iâd like to see is the initialisation - delete the xml and restart, or perform an initialisation so we can see whatâs happening.
Interesting stuff (unfortunately!). Give me a day or two to think about this one - itâs linked with the multi-instance-association configuration I thinkâŠ
thanks again. also i noticed, and i thing i have read a similar case somewhere, that after a restart the battery nodes are not getting fully initialized till the next wakeup. however they never get fully re-initialized automatically.
for example, after the restart my node 26 (fibaro motion sensor) is reporting all alarms without any issue but Battery state, temperature, ⊠are not getting updated. even after the 7200 sec wakeup interval not. actually i have to manually wake it up, which is OK as long as i have only one. but should that not happened automatically? in the snapshot binding the same was working fine.
Battery state is polled, so if the device isnât waking up, then it wonât get updated. However temperature should be sent unsolicited so shouldnât be related to wakeup.
The wakeup shouldnât be related to the binding - itâs a device function. Check the logs to see if itâs waking up - if it is, then letâs see what itâs doing.