I am using ethernet adapter on the new Homey 2023 pro.
Recently Homey 2023 started to drop communications over the ethernenet adapter. Happens from time to time, with no apparent pattern. When the issue occurs I can see in the network switch that no packets are received inbound on the ethernet interface from Homey. It is an adwanced switch. Resetting the switch or the port on the switch does nothing. But restarting Homey solves the issue. Due the the fact that this happens quite often, every two days, I have made a Watchdog flow that reboots Homey when it looses connectivity via IP as a temporary fix. But not a good one for longterm.
Anyone else experiencing this? I suspect this happened with the last firmware 10.0.0-rc.110, but not sure.
You can have configured Homey to communicate over both Wifi and Ethernet.
Anyway, I experienced only once Ethernet adapter being “frozen” / no traffic - I had to re-plug power to get it working back, did not happened since then.
Good tip. I looked into the DHCP server assignment and I saw two leases for devices named homey…something. It could be one lease for ethernet and one for wifi. But how do I turn off wifi? I thought that when one plug an ethernet adapter to the Homey pro it overrides wifi and “just works”. I cant remember reading or seeing anything about the need for turning off wifi. I configured wifi during initial setup. Was not my original plan, but the ethernet adapter did not work out-of-the-box. Had to update firmware via wifi first to get it to work. Also this problem occurred recently.
When Homey Pro 2023 goes offline when it stops communicating over ethernet. If it switches to wifi then the traffic is beeing blackholed/discarded.
I have opened a case in Athom support portal. In the meantime if anyone knows how to remove configured ssid’s from Homey Pro 2023 please let me know.
I’m seeing the same issue. Seen it on .110 stable, also seeing it on .115 unstable. Also using HP23 with the ethernet adapter, not using wifi. Yesterday enabled the sshd (not that silly ssh experiment) and waited for the HP23 to vanish from the network again. Happened early this morning, so was a couple of hour late to catch the actual logging leading up to a key even: dhcp lease expired, and unable to refresh. So my focus is now on what’s going on with dhcp, and/or, figure out if there’s a static ip option somewheres, before fixing that manually. Can tell you though, there’s not a single device on my network (60+) experiencing similar dhcp issues. To be continued.
Interesting (and probably something that Athom should know, I’m sure they’ll be interested in any logs; the same issue might be happening with the old Homey too since that also has a tendency to drop off the network at times).
What is the dhcp lease time? I am running 24hrs. Does extending the lease help? Or reducing it make it worse?
Normally when a DCHP lease expires, bouncing the client-facing interface triggers an update on the DHCP client. But not here. I do not know if the adapter passes the ethernet linkstate via USB to HP23, hence an update is not triggered.
I am wondering if there is a way to store Homey log files somewhere where it is reboot safe. Use a flow or something? Then I can grab logs when the error occurs, before the Watchdog reboots Homey.
Simply into timeline, it persists after restart.
Or, if you are hardcore user, you can use Simple log and log it into eg. Synology NAS or some Syslog server.
You could also use the SimpleLog for logging, and then write the log with the FTP client app to a FTP through flowcards.
Altho using the SimpleLog to connect to a Syslog server wil make sure all logs are contantly "exported’.
There’s an easy test for that: pull out the network cable and check if on Homey Developer Tools the ethernetConnected property is set to false (spoiler alert: it does).
I wonder what happens when the Ethernet DHCP lease expires while the WiFi lease is still valid.
I have my hp23 in a sep. vlan for all things IoT and currently lease time is set to 2 hrs. But the hp23 is extending per approx. 1 hour. (It’s managed by Debian’s NetworkManager.)
Link state looks just fine.
There are many options. I went with NewRelic for now, installed the infra agent and configured it to pick up on the various log files as found on the system. So that’s mostly systemd’s journal and logs from /user/log/*.log. The combination provides lots of metrics and logs in NewRelic. But you can use any platform that supports their agent running on linux on aarch64.
Tiny update, it doesn’t look directly related to dhcp lease expiring, my hp23 has survived many of the renewals (every 1 hr). I was focused on ipv4 though, but ipv6 seems to be working just fine as well. Next up would be, finding out what happens to the ipv6 interface when ipv4 vanishes again.
Just a quick input. The DHCP client will try to renew the lease via an unicast message to the DHCP server when half the lease time has gone. So if it renews the lease every 1hr when lease is 2hrs makes sence.
I have not seen the issue the last few days. Have not done any changes. I am not running dual stack in the network, just ip4.
Another small update; I’m now convinced it’s to do with the ethernet adapter. 3 or so days ago, it wouldn’t even come back online after a restart. It’d just act as if everything was fine, even obtain a lease, but then zero traffic. Let it sit for a while, I mean, have other things to do, no time. But then noticed, occasionally it’d restart by itself, must be some watchdog timing out then. Anyways, this Sunday decided to reconfigure the darn thing to wifi. Clearly it won’t handle running both wifi and ethernet correctly either, it prefers ethernet as soon as you plug the cable back in, but in my situation, then no traffic anymore. Decided to simply remove the ethernet adapter and have it run wifi for a bit. Well… It’s been running fine ever since. So either the adapter is crap, or something funky going on with drivers. Or… just a thought, nothing to back it up just yet, but what if the power supply just isn’t good enough? I remember there’s been trouble with the first batches, and then they started shipping the 2.1Amps. I have the 2.1A version, but maybe it’s not stable, but surely I’m not the only one experiencing these issues then or?
oh that’s interesting, I stand corrected. Somehow assumed it was about (insufficient) power, mea culpa. Now I do wonder though, maybe someone knows, how’s the ethernet adapter making any difference. I was reading this again: https://mailchi.mp/217770f9b300/homey-pros-power-adapter and notice the conditional suggestion of using a Raspberry power supply. That particular ps has 5.1V, and provides 3A. If the suggestion is, you could use that ps, just not when you also use the ethernet adapter, then the assumption is, there’s at least a voltage drop of 0.1 from the ethernet adapter to the homey (cm4)? Hence the need for a 5.2V supply? Guess I could give it another shot with the adapter but then carefully watch certain metrics, related to ps issues. Just can’t believe I’d be the only one unless I’d just have a faulty adapter. Another thing I’ll try, just connecting it to another linux machine with Debian 11, same driver, see how that goes. Hell, can even make that a spare Raspberry Pi so it’s same arch.
