Massive unexplained Z-Wave performance drop in the last couple days

My setup has never been super snappy but for the past year, has been reliable and fairly responsive. I have 44 Zwave devices, from newer Zwave+ to older regular Zwave devices. I use the Nortek Zwave+Zigbee stick, and am up-to-date on the latest zwave2mqtt/zwave-JS stacks.

What I’m seeing is hard to diagnose and maddeningly erratic. At times my devices (lights, locks, etc.) respond near instantly, other times they take up to 30-40 secs to respond, if they don’t completely fail to process commands.

I enabled debug-level logging and have been looking at timings, nothing weird came up in the logs, it’s like most commands get “stuck” and either go nowhere or take forever. I didn’t see any weird message storms or other unexpected traffic that would explain the dramatic reduction in performance.

I’m about to throw in the towel and get the Aeotec Stick 7 plus and Sonoff Zigbee 3.0 dongle and move to Zwave 700 stack, and completely rebuild my Zwave network one device at a time. But before I do that, I thought to check with the community and see if my issues are unique or if others are experiencing similar things.

Any ideas on what I could do to try and diagnose this further? Thanks!

This type of behavior happens if I have a device that is failing. Mostly this has been relays in light switches failing. Itt can also be when the z-wave dongle gets hung. When it’s the dongle I pull the dongle, reinsert and then reboot my computer. This has always worked. Just rebooting does not work.

3 Likes

Thank you @AllHailJ for the insight, you were spot on. The issue was caused by a new device I added a few days earlier, a Kwikset Home Connect 620 Z-wave enabled lock. Initially it worked fine, but apparently it went rogue a few days later and was causing the network issue I described. I didn’t initially think of it because I added other devices as well and never thought that a battery-operated, leaf-node device could cause the massive issues that were crippling my entire Zwave network.

It doesn’t make me feel great about the solidity and reliability of Zwave as a protocol, how a single leaf node could literally bring the entire network to its knees is beyond me. Are there ways to strengthen my Zwave network and make it more resilient to these types of issues in the future? I’ve done so many network repairs I lose count, not sure what else.

Perhaps get rid of and replace all non-Zwave+ devices? Anything else?

In any case, thanks much again!

What I think happens is the mesh network gets overloaded when a device gets stuck and is calling home too often. I have an automation that fails with lights clicking on and off when a node goes crazy. I have had z-wave with ~90 devices for a decade now. I have ~30 shellies, and they can get weird also. I am hopeful that MATTER will help fix some of this. Nothing will ever fix a failing device, however.

I would not replace the z-wave devices you have. All networks have problems.

1 Like

This specific device seems to be poorly implemented (not the only one!), and basically floods the network with requests, as discussed at Kwikset HC620 Lock Constant Zwave Chatter · zwave-js/node-zwave-js · Discussion #5566 · GitHub.

A workaround has been added in a newer Z-Wave JS. You can re-interview the device when your application software (add-on, container, etc.) is updated to include it.

1 Like

Thanks much for relaying this issue. Ugh… I will try that, though I added the lock about a week ago, and I’m always up-to-date with any zwave update on my HA instance… Will try again and report back, thanks!