Can't Delete Ghost Device

But .... if something passes z-wave certification, are the bugs real? /s

(tagging @mike.maxwell)

1 Like

bugs don't exist until they are observed...


I am not suggesting this problem with ghost devices is a bug, or that the devices didn't pass certification. I suspect that when Silicone Labs developed their software requirements, they did not include having a way to delete problem devices. I am suggesting that this design SUCKS, and Silicon Labs should change it.

Alternatively, Silicone Labs should provide a free Z-Stick with every Z-wave device so that they can be removed when there is a problem.

In my limited experience with such devices, the only solution unfortunately was a UZB stick. I got mine from Mouser and an very, very happy to have it around. :smiley:


That's most unfortunate. :frowning:

They did find it and address it.. There is a SDK update that is manual in 2.2.8, pushing for it to be automatic in 2.2.9..

They switched databases on the SDK from a proprietary format to SQLite.. The previous format was easily corruptible.. When this happens you get craziness because the nodeId in one place doesn’t match the nodeId in another. This mainly manifests when trying to remove a failed node.

Now that being said, the update does not retroactively fix nodes that already had database corruption, but it should prevent any new corruption from forming.


You missed the "/s" in my comment.

1 Like

And since I wrote that back in June, I have applied the SDK update (I'm a beta tester) that Bryan mentions above.

Things seem much more stable, but admittedly I haven't added or deleted very many devices, nor have I had ghosts to deal with. The few things I did add seemed to be fine though.

So I can't say things are solved, but I believe them to be much improved.

1 Like

UZB sticks do not work all the time. I have had some luck with the stick but it is not the complete solution that it is hyped up to be.

The only time I am aware of, is a node that responds to a ping.. You can’t remove a node that is online, by any method.

I have a melted node that is on the coffee table (not powered up) and I can not remove it with any of my UZB sticks. The z-sticks actually lock up on solid blue light on the stick. I have added and remove several other nodes after running into this problem and the stick work fine with them.

So bugs are like quantum particles????


Sooooo, its Hubitat that is the problem for not instantly implementing the revised SDK? :wink:

I do hope that the next release of Hubitat contains this revised SDK... before I have another ghost device.

@bcopeland, is this revised database type likely to result in fewer problems (database corruption) with unexpected power outages?

I didn't miss the "/s", Sorry, but I didn't know what it means.

That the preceding sentence was written in sarcasm

1 Like

Not exactly.. Great care had to go into this database update of live hubs everywhere.. Then there was also the debate on how prevalent the issue was in the wild.. 2.2.7 gave us an idea of how many people were affected with the additional remove device logging we added.. 2.2.8 became a staging release to test the update process on a wider audience..

Major changes like this have to be cautiously planned out.

It’s quite possible.. But I have seen the corruption on hubs that have never been off of UPS power too.

1 Like

Much better to proceed with caution when looking at something like this, as doing it wrong could mean losing your zwave table. Keep up the good work!


I assume this is something that will be in the release notes when it goes live, whether it's 2.2.9 or 2.2.10 or whatever? I just ordered a C7 to upgrade to, but don't want to start until this goes live. I'm not in any hurry tp upgrade atm.

You can install the new SDK on 2.2.8, in fact I recommend it, especially doing a new install

1 Like

Oh... ok. My C7 is supposed to arrive today so I'll check that out later. So the SDK can be updated separately from the HE firmware?

Download the Hubitat app