Open gjaegy opened 4 weeks ago
Dongle-E does not support hardware flow control, so using ember-nohw or main branch, it will be the same firmware. That ember-nohw branch is only relevant for adapters that support hardware flow control (Skyconnect, Yellow, etc).
The log you posted shows lixee trying to read from the device even though z2m hasn't started yet, so there is definitely something very wrong there. It's also taking way too long for messages to come back, based on those timestamps.
PS: The ZLinky is known to be a terrible spammer, which can cause major issues in already-unstable networks. You may want to try running a clean Z2M install without pairing the ZLinky, first, to see if the issue disappears._
Thanks a lot for your support, much appreciated.
I've already rebooted the Pi (power off) but that doesn't prevent the issue to appear again, sometimes quite quickly after reboot.
I'm going to try your suggestions one by one. I've now plug the dongle through a power usb hub, let's see if that helps.
In the case the issue persists, I'll then disconnect the ZLinky - it would be annoying if that was the cause as th'he Zlinky was the primary reason I set up HA 😂
The zlinky is in the same room as the Pi, so that part of the network shouldn't be that unstable (i'm not a zigbee expert but I understood it's a kind of mesh/P2P network - the ZLinky should be directly connected to the Pi dongle.
Understood about flow control - the amount of bad information one can find in internet forums is crazy... So I guess I can remove the 'rtscts' line in my config, is that correct (and install the main branch version of the driver)?
So I guess I can remove the 'rtscts' line in my config, is that correct (and install the main branch version of the driver)?
rtscts
defaults to false if not in the config, so yes, you can remove it, or leave it to false, same thing. You can keep the firmware til next update, since it's the same anyway.
As for the ZLinky, if you can't fix the problem with the suggestions, removing it from the network will at least give us an indication if it is the cause or not. Then we can go from there.
Can you provide the debug
log from right after a clean reboot of HAOS (not from after a Z2M crash/restart)?
It seems I haven't experienced any crash for the last 48 hours, which is clearly a record.
What I've done so far:
While I'm surpised there would have been any under-power issue (I've checked the Pi under-power extension installed in HA and it didn't report any low-power periods), another aspect of that change is that I've slightly move the antenna to make it more "visible", so maybe the reception is a bit better now (who knows).
I suspect the ZLinky to have been misconfigured with a high update rate, however, I initially increase the update rate when I bought it, but since it didn't change anything, I rolled back the change - possibly it wasn't applied correctly, hence the "reconfigure" operation fixed that ?
I don't know, really.
For information, before I made the two changes mentioned above, HA start process often failed when the PI was rebooted (it crashed before start was completed). To solve that, I had to unplug the dongle-e, reboot the Pi, then plug the dongle-e once fully started, and start Z2M at that point. Could that be a hint power was an issue ?
I would suggest to leave that ticket open another week or so, I will keep you updated and try something else if needed.
I'll send you the log you asked for (but possibly, start is OK now).
log.log Attached the log file generated after a fresh restart of HA. Let me know if that's what you asked for !
Hello,
I have the same behavior and i have a ZLinky too... After updated my dongle weeks ago to last stable version, the crashes happens less often. But 2 days ago the major part of my Zigbee devices was unavailable again... I think it is the ZLinky device which fall the first each time.
I will try to see if i can reconfigure it to reduce the update rate. For information, before using Zigbee2MQTT, i used Zigbee Home Automation without any problem (i made the change for a compatibility device).
@gjaegy, can you share your configuration for the ZLinky and the process to reconfigure it?
Hi Thomas,
I've rolled back to the default configuration, and simply increased the precision of the output to 3 digits.
However, Z2M crashed this morning, after 2 days without crashing. Attached the log.
The only change I can think of: I've put the Linky cover back on the device, hence covering the ZLinky under a thin 1-2mm sheet of plastic. Could that decrease the signal and cause the transmission to fail ?
So, my problem is still unsolved I'm affraid :/
OK, so we have the same configuration. My device is covered too but my RaspberryPi with HA is very close (under 30 centimeters). And to notice that the ZLinky is not the only device which doesn't provide data (some others, not all, can be unavailable in HA). When this happens, i don't succeed to find related logs...
If it can help, check out this thread, a few users had some past troubles with the ZLinky and tried various changes: https://github.com/Koenkk/zigbee2mqtt/issues/21198#issuecomment-1929344726
@thomasGomes Seems you have some other problem if devices are unavailable. Check your routers, see if one is causing trouble. You can use https://nerivec.github.io/z2m-ember-helper/ for a quick overview of your network based on a log file (the longer the runtime, the better the statistics).
@gjaegy Seems Z2M is forced to restart way too often because the ZLinky is bombarding it with retry messages for no reason (it eventually can't keep up), but the crash at the end appears unrelated:
[2024-06-14 08:46:40] error: z2m:mqtt: MQTT error: getaddrinfo EAI_AGAIN core-mosquitto
[2024-06-14 08:46:56] error: z2m:mqtt: Not connected to MQTT server!
Looks like some trouble on your other network (can't resolve the IP of the core-mosquitto
hostname).
Hmmm I see, weird that getaddrinfo() fails. MQTT server is running on the same device (the Raspberry Pi) within HA. Why would that call fail then, any idea ?
I’ll contact the ZLinky author, maybe he can fix that “spam” issue in the case there is an issue.
What happened?
Hi,
I'm having big troubles getting Z2M to work in a stable manner. I'm using HAOS on a Raspberry 3B+, along with a Sonoff Zigbee Dongle-E device.
I'm using the new "ember" Z2M driver.
I initially updated the firmware using a standard release, then, after some investigation, switched to the "ember-nohw" branch of the "https://github.com/darkxst/silabs-firmware-builder/tree/ember-nohw/firmware_builds/zbdonglee" repository (tried both 7.4.1 & 7.4.3) with hardware flow control disabled in the config (rtscts: false), but I can't get Z2M to run a full day without crashing - it always crash after some time (usually a few hours, sometimes a day, but that not often).
I'm really not sure what else I can try at this stage...
Below my current config, and the log for the latest crash today.
Thanks for any help !!
What did you expect to happen?
No response
How to reproduce it (minimal and precise)
No response
Zigbee2MQTT version
1.38.0-1
Adapter firmware version
7.4.1
Adapter
Sonoff Dongle-E
Setup
Raspberry 3B+ with HAOS (2024.6.1)
Debug log