C4 to C7 Network Migration - Resolved several Z-Wave issues with DNI clean-up

Hi folks. I'm relatively new here, forgive me if I'm repeating well known concepts, but I missed some important points in my migration process. I've seen a few other strings about issues with Z-Wave on a C-7 Device, and want to highlight my recent experiences in case they help others.

tl;dr: The documented Z-Wave network migration process does not identify the significant impact of an incomplete migration. A partial migration left defined "devices" with illegal Z-Wave DNIs, as well as partial radio associations. Cleaning these up resolved significant Z-Wave radio issues.

The detail:

I have been running a C-4 HE with about 45 Z-Wave devices, plus integration with Google Home, Hue, and a few oddball one offs. Over the (US) Thanksgiving break, I attempted a migration to a brand new C-7, using the recommended procedure from this forum.

While this was an awesome how-to guide, I think it could do a better job of calling out that partial migrations are likely to cause problems, and require additional clean-up.

In my case, several devices were left unmigrated due to technical or physical access issues. There were also a few false starts (Jasco / GE switches anyone?) that left "partial" Z-Wave Radio Devices in my Z-Wave Details with no associated device in the Devices tab.

Some operating symptoms after the migration:

Running "Repair Z-Wave" would crash my Z-Wave radio. Hard boot was required to get it back (thanks again to @672southmain for the help on getting the radio back up and working reliably: [2.2.4.156] Z-Wave offline)

Some devices simply couldn't communicate back to HE (this may have also been a bug in 2.2.4.156.)

And just generally... "weird" phenomenon like extremely slow Z-Wave commands, on the order of several seconds to respond.

Resolution: I thought I should make the system as clean as possible, so I started to delete devices that were not migrated. Then I ran into an error 500, and the place-holder devices wouldn't delete. I found I also couldn't remove incomplete radio associations from the Z-Wave Radio page.

There was some thrashing, but here's where I ended up, in this order:

  1. Change the DNIs back to a legitimate DNI (thanks to this post for that tip) to resolve the 500 errors. Explanation: "X" is not a valid character for a DNI, although the migration procedure recommended using "X" as a prefix for place-holder addresses. During migration this guarantees newly associated devices won't have a conflicting address. Unfortunately any left-over devices with X prefixes cause problems. Coincidently, "A" (also used in the procedure) is a legitimate DNI character and those devices delete, so it wasn't obvious to me that the "alpha" character was the issue. However DNIs are Hex addresses, so A-F are okay, but other (non-Hex) alphas are not. Change the X to any Hex character that won't cause a conflicting address with an existing device.

  2. Once the DNIs are legal, Delete the devices that don't actually exist from the Devices page (modify or delete apps that reference them if necessary)

  3. Finally, go to Settings > Z-Wave Details and on any remaining radio that seems incomplete (e.g., no "in: out:" nodes listed) click Refresh, wait for the refresh process to complete, then Remove.

I had to reboot once to ensure all radio associations were clear, but once they were, I shut down to red-light. Unplug. Wait 30 seconds. Restart.

Things looked promising. I went back to Z-Wave Details and pressed Repair Z-Wave. For the first time since upgrading, the C-7 whizzed through a total rebuild with no errors, and so far the rest of the house seems normalized again.

The improvement in C-7 performance, network reliability, and my general mood are all remarkable.

If you've followed the forum procedure for a recent migration, and your Z-Wave seems quirky, seriously go back and check for any stray DNIs and radios that didn't get cleaned up.

Hopefully future code will pick up some of these errors (and perhaps an advisory can be added to the migration procedure) but in the meantime I hope this is useful for a few folks.

And again, if this is old news, apologies to those who have already documented this path before me!

7 Likes

The instructions call for removing these devices immediately after each join. So yes, leaving them in the system will cause a problem.

1 Like