Seriously weird behavior with UI

[GrizzlyAK]

Hi guys, I've been seeing some weird behavior with the UI for some time now. I was hoping it would improve with updates, but it seems to have only gotten worse. This started before I upgraded my controller from ZStick 5+ to the ZWA-2. It used to work fine on this same system. For example, I've noticed the following consistently:

(1) When doing configuration on a particular node, I expand the node to access the Config parameters. I type in a value, click >, and it starts doing its thing. The node collapses and I have to repeat the process to open it back up. It won't stay open. Its like it is doing a random refresh.
(2) It is VERY slow and laggy. It didn't use to be. I noticed it about a year ago, but ignored it because I don't have to use it that much anymore. Recently, I had to re-add a shade remote that once was included, but had to exclude to add new shades. I have been unable, until today, to include it back, even trying a completely different remote control unit (brand new). I tried it MANY times and it always failed with basically that error message "inclusion failed". The UI would often show multiple entries for the inclusions all stuck on ProtocolInfo. I was able to add it back today using the HA Integration UI instead, first try.
(3) Also, to go along with the laggy behavior, sometimes the popups that appear when you hover the mouse over the Last Active, Status, Power, and other such columns, remain as you move off, In those instances, I can literally fill the screen with these popup messages by moving my mouse around, before the UI randomly refreshes.

I'm running it on Firefox v150.0.3 (latest) on Windows on the same LAN as my HA RPi. The UI used to be smooth, fast, and solid, but it is very hard to use these days. Is anyone else seeing this behavior or am I just the lucky one? If I'm just lucky, do you have any advice to offer that might remedy these issues? Thanks.

[robertsLando]

Hi @GrizzlyAK. A screen record of the issue + app/driver logs would help a lot here. How many devices do you have?

[GrizzlyAK]

Hi Daniel,

I've including a screenshot of the lasting popups below. Note that this happens regularly for me. Sometimes they work as expected, but other times, they stack up as I mouse over. It's almost like something is running in the background that is preventing a screen refresh, because at some point after this happens, they all just disappear, as if the screen refreshed. I have 94 devices. Logs are attached. Please let me know if you need anything else. Cheers. Shane

zwave-js-ui-store.zip

[robertsLando]

Hi @GrizzlyAK, thanks a lot for the detailed report and the logs — they were very useful. I dug into all three symptoms you described and they line up pretty cleanly with what's in the code. Quick rundown of what I found and what we're doing about it:

Why the config panel collapses when you press >. The store was both mutating the node object and re-splicing it into the nodes array on every update — that array-level pulse was invalidating Vuetify's expanded-row state and dropping your edit. With 94 nodes throwing constant updates this happens a lot.

Why the tooltips stack and stay on screen. The tooltips on Last Active / Status / Power were using the v-tooltip directive form. When the cell underneath re-renders mid-hover (which happens constantly during a value-update burst), the directive leaks the overlay — that's the "fill the screen with popups" effect you can see. Eventually a bigger re-render tears them down all at once, which is the "screen refresh" you noticed.

Why "Inclusion failed" leaves stuck entries on ProtocolInfo. When you start inclusion the controller assigns a NodeID immediately, even before the device sends its node info. We were correctly showing that partial node in the UI, but on inclusion failed we weren't cleaning it up — and zwave-js doesn't always emit node removed for inclusions that abort mid-protocol. So they just piled up. The HA integration UI worked first try because it calls removeFailedNode for you.

One extra thing your logs surfaced. Of ~38k value updates in the 20-hour window you sent, about half are no-op refreshes (the device re-reports the same value, like 100 => 100 or 0 => 0). Each one was triggering a UI re-render even though nothing visible changed. With your number of devices that alone explains a lot of the lag.

We're working on a fix in #4649 that addresses all four points:

• Stop broadcasting no-op value updates to the UI socket (MQTT/HASS still receive every refresh — only the UI is gated).
• Stop the redundant array splice in the store so the expanded row stays put while you edit.
• Switch the tooltips to the component form with a small open-delay so they can't orphan during re-render bursts.
• Clean up ghost nodes after a failed inclusion so they don't sit at ProtocolInfo forever.

While you wait for the fix, a couple of things to try that might help right now:

1. Run a route rebuild on the controller. Your logs show 18 "Controller is unable to transmit" cycles in 20 hours and one full driver recovery failure. That's a sign of marginal links in the mesh — possibly from the ZStick 5+ → ZWA-2 swap. The recovery cycles each end with a burst of queued values (we saw bursts up to 45/sec), which is when the UI is most likely to misbehave.
2. Spot-check the chattiest nodes: 066, 012, 026, 019, 135 are each firing ~3,600 updates/day. If any of them are misconfigured to report unchanged values aggressively (some Aeotec/HomeSeer sensors do this by default), trimming their report intervals will help even after the fix lands.

Will close this out once the PR is merged and released. Thanks again for the great bug report!

Cheers,
Daniel

[GrizzlyAK]

Hey, sorry for the delay in responding. I did read this after you responded, but failed to act on t6he suggestions. I've looked at the chatty nodes you listed and one is my office motion detector, and the others are Zooz Zen15 Power Cords that I have attached to fridge and freezers that I monitor frequently for outages, device failures, etc., so the chattiness is expected. I may be able to trim them a bit and will take a look. I just updated ZWave JS UI, and it looks like it updated your UI as well. I'll let you know if I see any remaining issue as time goes. I appreciate your efforts on this. If you'd like me to check anything in particular, let me know. Cheers.

[robertsLando]

The fix has been merged so all your issues should be sorted out now, let me know 🙏🏼