Zigbee Mesh Troubleshooting

[ritchierich]

From time to time community posts appear where a hub's Zigbee mesh goes down the toilet so I wanted to create a post about what I did to solve an issue that appeared the past few days.

I have a strong Zigbee mesh of comprised of 99 devices both mains powered switches and outlets and many battery based such as leak, contact, and motion sensors. Yesterday I noticed a few weird cases of lights not turning on or off as they should based on rules I have setup within HE but unfortunately this does happen from time to time so I just let it go. This morning things got a bit worse and I saw a few battery devices hadn't reported their battery level in the past 24 hours. So I reset them and attempted to re-paired them and a few of them wouldn't pair so I knew something was up with my mesh.

I came back to my computer opened the Zigbee logs to see if something was spamming the mesh - Settings \ Zigbee Details \ View logs. I wanted to give it some time to collect data so I left this tab open for several minutes and came back and then clicked on each device to visually see its log entries. One of them had a ton more items than the others.

As @bobbyD frequently posts, a single device can destroy your mesh and that was the case for me. After several minutes I saw one device had created several hundred log entries. The device was an Iris v2 Contact sensor on one of my doors. Inspecting the device details, battery was reporting 62% but I wanted to doublecheck it with my battery tester and low and behold it was almost dead. I installed a new one and boom everything was working well again. The devices that were refusing to pair earlier this morning paired right up.

So lessons learned:

• Don't trust battery levels reported by devices and have a good battery tester available to get a good reading. For example I have leak sensors that will report 0% for 6 months or more before the battery finally dies.
• Utilize the tools available to monitor mesh issues. The Zigbee and Zwave logs are great but you can also export this data to a MySQL database using NodeRed connected to the HE websockets if you have those setup.
• Devices can do strange things when the battery levels get low. I have many Iris v2 contact sensors and I don't remember one spamming the mesh like this situation.

[user3420]

That's interesting. Thank you for that information. I too have a network that becomes VERY laggy, and some devices very slow to respond. Now I know what I need to look for. What would also be nice is to have a network alert to let you know that a device to clogging up bandwidth, like the zwave/zigbee crash, low memory, etc alert options. Perhaps create a device to handle this? Thks.

[dnickel]

I use the, not suported, Health Status, as a good catch-all.
It is true Battery is a guess; what device updates it's battery level for you just before it dies?
I have many devices and 7 mains powered ZB repeaters strategically placed but yesterday 6 devices popped on to the Device Activity List as Offline; I had to re-discover these.
Zigbee Msgs is your friend for sure.

I can only attibute these fluctuations to Wifi/RF bursts.
I have changed channels a few times looking for stability but it's usually the same result.
My hub lives a sheltered life according to the scan.

[tray_e]

@ritchierich

Devices can do strange things when the battery levels get low. I have many Iris v2 contact sensors and I don't remember one spamming the mesh like this situation.

I have the samsung contact sensors from back in my smartthings days. One in particular, on my front door will start oscillating between open and closed when the battery gets low. (Can be REALLY annoying if it happens to hit the critical level with HSM in sleep mode and you have intrusion alerts set up..... ask me how I know )

And for the record, the first indication was wonky mesh issues with things / actions being delayed, etc. Now, if things start getting a little slow, I start checking battery dates I have recorded in each device.