There were updates to Z Wave functionality in 0.58. These updates have caused me great grief lately. What was working is now broken and there’s no explanation. I’ve pretty much ripped my Z Wave configuration apart, started with a clean slate and empty zwcfg xml file, excluded all nodes and added them back in one at a time. Nodes I try to add as secure (Aeotec Smart Switch 6) fail and/or have random and invalid configurations loaded for them. My switches and dimmers fail to operate properly one minute, but then work just fine the next for no rhyme or reason. Status updates work when interacting with the switch or light from HA frontend, but don’t update status when physically interacted with. Then I try again about 5 minutes later and they start reporting their status just fine. All of the nodes show “Ready” and show Node 1 (Z Wave Z Stick) as their first neighbor.
It just doesn’t make sense. The next logical conclusion is the updates to the Z Wave code in 0.58 broke Z Wave. I know we are on the eve of a new release, but I really hope these issues have been reported and resolved or code has been reverted or updated (I saw Open Z Wave has updated their library on GitHub recently). I can’t be the only one having pretty severe intermittent issues…
OpenZWave was last updated in 0.55, with the inclusion of python_openzwave 0.4.0.35. There have been no further releases of python_openzwave since, and no further updates to HA as a result.
Hey @Tinkerer, I’m not disagreeing with you. The base code (openzwave) may not have been updated, but there were improvements made and the Z Wave code has been touched:
I have a z wave switch that turns on multiple z wave lights if toggled. The lights are not physically wired to the switch. Last night I could toggle the switch both physically and in HA to my heart’s content and all the lights would turn on/off. This morning with absolutely no change in configuration, no reboots, no errors, the lights do not respond to the switch being physically toggled. They only respond to HA when I toggle the switch on the front end. Then I waited about 3 minutes while I typed this out and try the switch again and all of a sudden after multiple attempts at physically toggling the switch, HA starts seeing the status updates.
This is not a case of polling or not polling (see my other recent thread where I discuss this and my findings). The switch I have is perfectly capable of sending its status updates to HA. There is something else going on where HA is not getting these updates. There is nothing I can see in either the HA log or OZW log with an error for z wave. I have the logging level for Z Wave set to Debug.
EDIT: After every state change either physical or from HA, the OZW log always has this:
2017-12-02 15:47:40.288 Warning, Exception: Manager.cpp:2499 - 102 - ValueID passed to GetValueFloatPrecision is not a Decimal Value
2017-12-02 15:47:40.288 Warning, Exception: Manager.cpp:2499 - 102 - ValueID passed to GetValueFloatPrecision is not a Decimal Value
I came across this thread after noticing strange issues with my Fibaro door sensors. Have 5 of them, only 1 is working, where they were all flawless for more than a year (flawless Fibaro only network by the way, with a Aeotec Gen5 usb stick). Didn’t change anything but upgrade to 0.58.1 from 0.57.2. Led’s are flashing on triggering the sensor. Didn’t have time to look further into it, but the last report for 4 of the sensors is half an hour after the upgrade. Very strange. Going to dig further.
/edit
Downgraded to 0.57.3, no dice, sensors not reporting, nothing showing up in the OZWlog.
/edit 2
Restored snapshot from before upgrade of 0.57.2 > 0.58.1. Same behaviour. Next up, battery replacement. Replaced a battery of one sensor, presto, it worked. Upgrade to 0.58.1, still works. So no flakey behaviour here, but a very big coincidence of 4 out of 5 door sensors stopping to report due to low battery on the same evening. Very very strange. Will replace all other batteries now.
/edit 3
All my sensors working flawlessly again after battery replacement. Think it has something to do with the cold here. All batteries were very cold. All sensors which needed replacement are located on an outer door.
Battery devices behave differently than mains powered devices. In my experience, HA is having trouble handling all of my mains powered devices. It’s just flat out broken. I spent all day removing all nodes and adding one at a time and HA worked with 1 switch added. After I added more, it stopped working. My final conclusion and stance is HA has a broken implementation of OZW. I’m happy to post my config, logs, and any other troubleshooting results, but unless someone shows me otherwise, I conclude it is broken.
There would be an absolute flood of Zwave related posts here if there was a fundamental problem with the HA Zwave implementation. There isn’t. There may be some specific device or combination that presents a problem in your situation, but I think you will need to track that down. You will need to go through the OZW_Log and home-assistant log files to see what is happening or not happening. If there were more users with an issue it certainly is easier to track down issues, but this seems like a somewhat unique problem.
I would try a fresh install and not add your current configuration. Just add the zwave entry to the fresh configuration.yaml file. That will give you a baseline to test from.
I run a docker container, so I can’t change anything about the environment or base code. I wiped my z wave config, removed all nodes, added them back in one at a time, and did that about 3 times so far with no change. The problem is my z stick gives me feedback when a device changes, but HA doesn’t see that. I have the z stick less than 5 feet from a switch. I toggle that switch and I see the blue light on the z stick flicker as the two communicate, but I check the OZW log and I check HA and there is no indication any communication took place. There’s no recent time stamped communication with the node, there’s no change to the switch state in HA, and there’s definitely no error logs generated. But if I even toggle it off and back on one more time, sometimes, but not always, then OZW and HA will show communication in the log.
So I did a test, I started a docker container of just Open Z Wave Control Panel and watched states in there. They update almost instantly (give or take a couple seconds for the webpage frontend to refresh) and it shows node communication happening, the first time, every time.
As for other components causing issues with the z wave component in HA, well I’d say that’s a bug or problem that was missed or inadvertently introduced. I have disabled a few platforms recently, but in general, my configuration has not changed for the past 2 or so months, except for the occasional group, script, or automation updates.
Again, if this was a generic problem everyone including me would be reporting your same issues. They aren’t.so it comes down to you figuring out the issue. You may have some unique condition that causes the problem and unless you can track it down so someone can fix it, you are stuck. Try going back a version. Still an issue?? Go back two versions. Still an issue? Sadly in any opensource project, you are the first line tech support department. Try installing hasbian??
Yep, I realize I’m pretty much on my own for troubleshooting. I did some troubleshooting, did some research, did some more troubleshooting, tried reaching out for help, then came to my conclusion.
Now, with that being said, I want to apologize to the community. I’m sorry for jumping to conclusions. No, OZW is not broken. I was wrong. My understanding of Z Wave and the OZWCP implementation is incomplete. I made a false conclusion based on partial knowledge and research.
What I was experiencing that caused me to blame OZW or HA was actually caused by my devices in a Z Wave mesh network. OZW was only doing what it was told and my devices were not communicating properly.
I’m going to cross post the rest of my findings in my other thread about polling devices. That will have my further findings.