Z-Wave Mesh seems to be deteriorating -- Suggestions on troubleshooting steps?

I've been running Hubitat C7 for more than a year with mixed, but mostly acceptable results. Recently I've noticed that some of my scheduled events are not occurring, like the Family lights not coming on at Sunset as the rule says it should. Although it is within almost line-of-sight and about 25 feet from the hub, it shows only modest signal strength and many route changes:

A peek at the topology suggests that the mesh is very weak despite many, many powered Leviton devices that could/should be repeating.

Strangely, the bottom two devices, Ring repeaters, which are about 25 feed from the hub (down a hallway, thus having several layers of drywall between them) are experiencing numerous route changes which seem to diminish their purpose.

Equally confusing is why an Aeon battery powered window sensor just 5 feet away from hub, which is rarely opened but is in line of sight from the hub, takes multiple hops to get connected.

A quick look at the log stats suggests that the system is not very busy.

Would appreciate any suggestions as to what I should do to troubleshoot?

Your approach looks like the correct one.. Have you installed "Hubitat Z-Wave Mesh Details" community app yet? It reformats the existing info to give you a much better picture of your devices and their neighbors. Helps figure out if a device is starting to fail possibly..

Also @thebearmay's hub information driver as well.. can give you insights into the hub like thermal and cpu performance.

40 kbps is perfectly within spec so should not necessarily be an issue but I agree especially for devices closer to the hub - should be at 100 kbps but your environment certainly plays a factor.

In your Z-Wave details list do you have any ghost devices - entries in your Z-Wave details table that look weird?

Do you have any oddball proprietary RF (900 MHz range) controller type stuff (security systems, remotes etc) or old cordless phones (remember those?) that could be causing interference?

Here's a fun rabbit hole to consider..

3 Likes

Thanks. I have the Z-Wave Mesh Tool loaded, but haven't used it to review this situation yet. Will do so over the weekend.

Yes...particularly if they lack any routing info in the last column...that would be a problem that you'd need to fix.

3 Likes

I recently crashed my well performing z-wave mesh with an ill conceived rule that was changing the level of a switch every 0.3 seconds that I managed not to stop when it reached the desired level. Since it didn't work I did not notice that it was continuing to run. Over the course of a couple of days everything slowed to a crawl. When I looked at the hub there was a cpu overload message, the database was full and the logs showed that z-wave had crashed. I deleted the rule. The mesh was a mess. In the end I shut down the hub, removed power for thirty minutes and it came back healthy. Several z-wave repairs before I shut down the hub did nothing that I could determine. Maybe restore to a pre-slowdown backup in case something has been corrupted.

@wfoley post a complete pic of your z-wave details page for us so we can see everything

2 Likes

Pretty ugly.



And, pretty hard to read -- sorry.

Ghosts are in the house...

2022-02-04 11_27_31-Z-Wave Mesh seems to be deteriorating -- Suggestions on troubleshooting steps_ -
2022-02-04 11_27_38-Z-Wave Mesh seems to be deteriorating -- Suggestions on troubleshooting steps_ -
2022-02-04 11_28_25-Z-Wave Mesh seems to be deteriorating -- Suggestions on troubleshooting steps_ -

Assuming this one is a battery operated contact sensor, it's OK if it doesn't show routing.

You you can try using the Repair button on the three light/light switches to see if that brings back routing. If not then you're going to have to remove those ghosts to get things back in shape. So try using Repair first. Use the Repair button on each switch, not the Repair option at the top of the page.

2022-02-04 11_31_20-Z-Wave Mesh seems to be deteriorating -- Suggestions on troubleshooting steps_ -

1 Like

If Repair doesn't work...you'll have to remove those ghost devices w/out routing and then re-add them to your hub.

The actual removal of the switches w/out routing could be pretty simple if you get lucky. You can try excluding the switches via the normal process (hub in exclude mode, action on switch to put switch in exclude mode, and device is excluded). But that may not work in this case.

If above doesn't work then:

  1. Kill power to the each switch (hit a breaker, or pull the air-gap on the switch)
  2. Hit the Refresh button on the Z-Wave Details page for that device (it may take more than once) until the "Remove" button appears.
  3. Hit the Remove button (also could take a few tries) to remove the device

Then look for the Device page for that switch in the Device tab list - it may still be there, and use the red Remove button at the bottom of the page to remove the swtich from your Device list.

Reboot your hub, and then after it boots up and settles in, add the switches again, making sure they join normally and show routing. To make things simpler and easier on your mesh, if a Security dialog appears during adding the switch, hit Cancel or Skip so the switch can join w/out security, simpler and easier.

To make things easier on you check the "In use hy" section for each switch so you know which automations they are being used in, so you can insert your new entries for the switches back into those automations.

3 Likes

@wfoley Ghosts.... As @danabw points out they need removed. They are crushing your mesh.

Shut down the power to the devices that made the ghosts. Shut down the hub and unplug at the wall (not the hub) for 20 mins. Power on the hub. Click refresh on those lines until the remove option appears. Remove. If that doesn't immediately remove it, wait a minute or so for the Force Remove popup.
0x55, 0x6d, 0x71, 0x79 are your ghosts. After you remove them, repeat the shutdown and unplug procedure outlined above again. Let your mesh settle for a few hours and see how thing are.

3 Likes

Working on it, guys. Had to Exclude and Include Steve light to get it to work. The Back landscape lights started working after a few tries (this is at the edge of my property and far from a repeater). Put new batteries in two sensors to get them back. Still looking for Jen's light module that was apparently removed from the AC plug. Appreciate the coaching. Didn't realize that these "empty" routings indicated "ghost" devices and would cobb up the network so badly. Heading out for the weekend, but will report back next week. Thanks, again. Nice to have you guys there for us.

4 Likes

Actually I'm here for the high pay and superb health benefits. Plus @rlithgow1 throws a BBQ every couple of weeks that's just the bomb. :wink:

Glad things are headed in the right direction, and that I'm not the only one who's family seems to take losing every portable IoT device as their life's mission. :smiley:

1 Like

Hey now, don't forget about the meth.... Lots and lots of meth.....

1 Like

Found the missing Jen light module and plugged it back in -- my wife had put it in the drawer!

Rebooted and left the system untouched overnight. Wow, what a difference! Still lots of red squares on the hub topology, but it looks like performance has returned to normal, at least as far as the RTT times and route changes indicate. Obviously, one of the "ghost" devices was causing the problem, but I still don't understand why the hub statistics don't give me a clue as to which one is consuming bandwidth.

Anyway, many thanks.

2 Likes

Woot! :smiley:

Glad you Sherlocked where the light module was and things are returning to normal ...we have SO much stuff that we've put away and now have no idea where it is. :wink:

So all the ghost devices are gone?

1 Like

Also there are 3 devices using S0. It's normally advised to exclude and re include those devices with no security.

Keypads, Locks, and sirens will normally not include or function correctly without some type of security - use S2 if you can with these, else S0.

4 Likes

As it turns out, two of the S0 devices are Kwikset locks that were pre Z-Wave PLUS devices that would only include at S0. LIkewise, the Aeotec Siren is also pre Z-Wave PLUS. All three of these devices were installed early in the transition to Hubitat (from Vera) and don't seem to have been a problem. As I review the events of the recent past, I believe the problem was Steve lights, a Leviton DZ6HD that was still functioning as a dimmable switch, but had lost contract with Hubitat. Trying to reset it with the air gap offered no relief so I had to exclude and re-include it.

All has been stable for the past three days. Strange -- maybe it was generating a bunch of undetected noise in the mesh? Unfortunately, having excluded it from the system, I no longer have access to the logs. Always a learning experience with Z-Wave.

3 Likes

I wonder if you can replace the lock modules with upgraded versions maybe even Zigbee?