Cumulus Support

I am seeing this line in my logs, on Cumulus MX 3.28.6 b3283, on a pi 4b. I can't see any obvious issues, but I haven't had a deep dive into my data. My own processing scripts that I run monthly on the day and month files have been processing without any hitch. I only came across it when troubleshooting some network issues (more on that below).

EDIT - I am co-running an install of CumulusMX v4 on a second Pi, with the same everything (apart from the Pi) and the error doesn't appear in the logs for this instance.

Do I need to worry, or fix something?
I have attached some logs and service consoles for context - there are several files because I updated and did lots of restarts to see if I could resolve it. Without luck.


Secondly, on the network issue - can anyone confirm if the Ecowitt GW1100A (V2.3.x) has inherent network problems? I've always had random missed pings, disconnections, invalid gateways etc - but never lasting for more than a few minutes. I had put it down to some quirks with my LAN, because my Homebridge and a few other connected devices had similar. Some troubleshooting prompted me to replace my network switch - which fixed 100% of all the other errors - apart from the Ecowitt. There is nothing else I can think of apart from a flaky wifi chip/software in the gateway.

Re the network issues. My GW1100 is fine but my GW1200 seems to have intermittent disconnects briefly. Is your GW1100 accessed by both CMX instances? All my gateways are only accessed by one CMX instance.

Stuart

whitling2k wrote: ↑Sat 07 Sep 2024 4:44 pm Secondly, on the network issue - can anyone confirm if the Ecowitt GW1100A (V2.3.x) has inherent network problems? I've always had random missed pings, disconnections, invalid gateways etc - but never lasting for more than a few minutes. I had put it down to some quirks with my LAN, because my Homebridge and a few other connected devices had similar. Some troubleshooting prompted me to replace my network switch - which fixed 100% of all the other errors - apart from the Ecowitt. There is nothing else I can think of apart from a flaky wifi chip/software in the gateway.

My GW1100A is running with firmware V2.2.7, and it does not have any stability issues with my Asus RT-AX86U. I have LibreNMS polling all my devices, it does 3 pings every 5 minutes and also checks that http response is ok. Availability of the GW1100 is 100% for several months. Of course some short glitches can be missed because of the 5min polling interval, but if there were many recurrences, they should appear in the monitoring graphs.

One thing to note. If you are living in congested area, the 2,4Ghz wifi is very likely to have all sorts of issues due to disturbance from neighbours. If that is the case, all you can do is selecting the wifi channel with least disturbance. Only channels 1, 6 and 11 are completely non-overlapping, but if someone has configured their router eg. to ch3, then both 1 and 6 are also ”broken” and suffer from disturbance.

At the moment, both instances are reading the same gateway, but it's the same if one, two or more cmx instances access it. It just disappears sometimes. Weirdly though, sometimes, it seems accessible, but denies access. It has always behaved like this - and the issue comes and goes.

My HomeBridge had similar errors - but they seemed to be resolved with the new switch. It's unlikely to be my wifi modem-router either, as I replaced that recently as well.

We're not in a congested area as far as Wifi goes. We live in a sparsely populated area with open woodland on one side, and low density housing elsewhere. I have one Wifi network (well, 2 - 2.5ghz and 5ghz), our neighbour has one and that's it. It's very rare I can pick up more than three networks when scanning.

I do have lots of Zigbee devices - I wonder if they interfere with Wifi at all? I do have issues with my Zigbee stuff actually, my Hue switches sometimes can't communicate with the hub, even when used ~30cm from it. Although my boiler and smart meters seem to have solid connections.

Perhaps I need to find a way of scanning for zigbee congestion as well as Wifi?

whitling2k wrote: ↑Sat 07 Sep 2024 8:03 pm I do have lots of Zigbee devices - I wonder if they interfere with Wifi at all? I do have issues with my Zigbee stuff actually, my Hue switches sometimes can't communicate with the hub, even when used ~30cm from it. Although my boiler and smart meters seem to have solid connections.

Perhaps I need to find a way of scanning for zigbee congestion as well as Wifi?

My knowledge about zigbee is a bit limited, but yes, it does share the same 2.4Ghz frequency range with wifi, and the channel numbers do not match directly with wifi channels. The easiest way to cope with it is to select non-overlapping channels for both. Here is one article regarding the zigbee and wifi coexistence. https://www.metageek.com/training/resou ... existence/

I've just moved all my wifi and (moveable) zigbee channels around. Thankfully, all my neighbours wifi devices are on channel 1, and they are all 2.4ghz - so a nice easy job to avoid. Let's see if it makes any improvement!

Now, for that pesky error message

That error message is *probably* Mx trying to store the sensor reception data.

If you turn on debug and data logging, and post another log file after the error occurs again I may be able to use that to reproduce the problem.

Thanks Mark, will post back in a few days.

Hi Mark - more detailed logs as discussed.

OK, the error is caused by the addition of the WS85 sensor by Ecowitt. Version 3 of MX does not decode this, it was added in to v4.

Perfect; I’ll ignore it then, I’m still dual-running MX 3&4 on different Pis because my 5b misbehaves on the scheduled reboots.