Error that causes loss of connectivity?

Configuration
1

Hi all, for a while now I have been experiencing quite rare network problems with my HA instance. This happens approximately once every two weeks, but I haven't kept track of exact timings. The only way to re-establish internet connectivity that I have found is to restart the host, 'network reload' in the CLI doesn't help. I've tried a few things like making the DHCP lease infinite but that hasn't seemed to help.

Some technical details:

  • HA runs in a virtual machine on a Dell laptop, the host OS is Linux Mint, the hypervisor is Virtualbox
  • The laptop is connected via ethernet to my router, which runs OpenWRT
  • I am currently on HaOS 18 beta, but this has happened in earlier versions too

It happened again today, so I decided to check the HA host log. I found this sequence of events:
First, even when nothing is wrong, I still see a set of entries like this, which occur approximately every 2 hrs. (VeulenSmart is the name of my instance)

2026-06-19 15:05:11.750 VeulenSmart systemd-timesyncd[535]: Network configuration changed, trying to establish connection.

2026-06-19 15:05:11.758 VeulenSmart systemd-timesyncd[535]: Contacted time server 162.159.200.123:123 (time.cloudflare.com).

2026-06-19 15:05:12.061 VeulenSmart systemd[1]: Started Time & Date Service.

2026-06-19 15:05:41.734 VeulenSmart systemd[1]: systemd-hostnamed.service: Deactivated successfully.

2026-06-19 15:05:42.104 VeulenSmart systemd[1]: systemd-timedated.service: Deactivated successfully.

I do not know if it is normal for the network configuration to change all the time, but at least everything works. Then, this error occurs:

2026-06-19 15:11:43.785 VeulenSmart dockerd[553]: time="2026-06-19T17:11:43.777220656+02:00" level=error msg="failed to validate image signature" error="resolving signature chain for image sha256:016716ae746e8612fb7ab1be0e88d83e3b7f5dada3f418288a4ef0c3bfcf9a7c: expected image index descriptor, got application/vnd.docker.distribution.manifest.v2+json"

I have no clue what that means, and it doesn't seem to relate directly to the network problem, but it's the only other error I see in the log. The next entries repeat over and over, and I assume they simply mean: "help, I can't find the internet". The timing of these entries also correspond to the time when connections to other (local and internet) entities disappear.

2026-06-19 16:21:12.361 VeulenSmart systemd-timesyncd[535]: Timed out waiting for reply from 162.159.200.123:123 (time.cloudflare.com).

2026-06-19 16:21:22.621 VeulenSmart systemd-timesyncd[535]: Timed out waiting for reply from 162.159.200.1:123 (time.cloudflare.com).

2026-06-19 16:21:52.967 VeulenSmart dockerd[553]: time="2026-06-19T18:21:52.967266026+02:00" level=error msg="[resolver] failed to query external DNS server" client-addr="udp:192.168.100.26:47876" dns-server="udp:192.168.100.1:53" error="read udp 192.168.100.26:47876->192.168.100.1:53: i/o timeout" question=";o427061.ingest.sentry.io.\tIN\t A"

Can anybody tell me

  • What the error means
  • Whether it can have anything to do with loss of connectivity?
  • How to solve these problems?
    Unfortunately, there are some devices spamming my router log, so I can't scroll back far enough to read what is there at the time of the error.

*small edits to clarify the problem more

2

2 hours discrepancy … not good

3

Can you run something like ntpdate -q time.cloudflare.com on the host? If not, then it could be an OpenWRT firewall or similar.

4

But those two hours are just the time zone difference with UTC, no? There's always 2 hr difference even if there is no error.

5

That would be a weird firewall that only gets active once every two weeks. The time server is not the problem, it's a symptom, it's just the thing that ends up in the log file when this problem occurs. The whole instance drops off the network, it cannot even be pinged or connect to local ESPHome units. Homeassistant still runs fine otherwise, Zigbee works, automations that don't need LAN or Wifi still work, etc.

6

Right, and What you get from cloudflare does not ”linger” with what your system says

Are you sure you have configured your timezone corret

7

Yes, my timezone is correct and my instance sees the correct time for automations etc.

8

image
image950Ă—278 19.7 KB

9

Then do the test as Inventor7777 suggest, and also check your overall network configuration/connection, DNS etc

It’s very hard to “assist” here as only you know how your network works and which APPs etc you have running

NOTE: I use “server time zone”

10

The 'ntpdate' command doesn't seem to exist in the HaOS shell, not sure how to execute it via ssh.

11

I have been digging like crazy for a long time in my network configuration, it was one of the reasons to go to OpenWRT, so I could at least have control. But I haven't been able to find yet what the cause is. I went back and forth between DHCP (with reserved IP) and static IP, tried switching everything related to IPv6 off, etc, etc. It's just hard to diagnose a problem that occurs so infrequently with a router log file that only covers two hours. I did at least find out just a moment ago how to change that.

12

How old is the Dell laptop? Go into the BIOS and check the date/time in there. It's a stretch, but possibly could be a CMOS issue on the hardware and if the MOBO has a CMOS battery, the battery could be dead.

13

Let's not go on a wild goose chase. There is nothing wrong with the time on any of my systems, I'm in UTC+2, that's all. The laptop's Linux Mint remains connected when this occurs, it's only the VM that gets disconnected. Could be a Virtualbox issue too but a recent upgrade from 7.0 to 7.2 didn't fix it either.

14

Aye, just a suggestion :man_shrugging: Weird obscure things can happen and I tend to look for things that people check last, lol

15

Hey, I should have checked what HA has by default.

Try running ntpd -w -p time.cloudflare.com and hit Ctrl C if/when it syncs successfully.

Have you assigned a static IP in OpenWRT or in VirtualBox or in HA's network settings? You said that the host Linux Mint stays connected, that honestly points to a VM connectivity issue. I can't help much with VMs as I use HA in bare metal as I don't want to fiddle with such errors.

If it is just the VM, you could try searching the web or asking AI if this is a known issue with VirtualBox. (e.g a setting)

16

Sorry for my grumpy sounding reactions, they weren't meant that way. You are all just trying to help, thanks for that!

When I run the ntpd, it syncs with a quarter second offset and a few ms delay. As I said, the time is not the problem, the time server error is just the symptom of not having internet connection, and the 2 hr offset between local time and UTC is expected because of my time zone.

I agree that it must be something in either the router, or HA, or Virtualbox. I'm doubtful it is the router since all other devices work fine. It could be Virtualbox, I will investigate a bit more into that. But it could also be something in HA, which is why I asked here.

I specifically was wondering if there could be a connection with the docker error, but looking at the timings I don't think so, there's more than an hour between the loss of connection and the error.

At the moment I am not using static IP, but DHCP with a reserved IP. I tried static in the past to solve a different problem, that didn't help then but it was a different configuration. I will try again and see if it works. The feedback loop is very slow though, it will take weeks before I know if it helped.
Virtualbox does not have any IP settings in it that I could find, from what I see it just passes through a connection on the host to the guest, spoofiing it as a specific device, the rest is handled by the guest OS (HAOS). The naming is confusing btw, what HA calls the host, is the guest to Virtualbox :slight_smile:

17

OMG, I both love and hate the fact that sometimes AI is a lot more useful than humans....
With the help of ChatGPT I completely diagnosed the problem. What happens is the following:

  1. The slightly flaky usb-c to usb-a adapter connected to the usb-to-ethernet adapter briefly comes loose and fixes again because I touch something nearby
  2. After the plug-unplug of the usb quickly recovers the network on the Linux host, but it throws off the Virtualbox network emulation, which causes HA to think it still has an ethernet connection but it effectively doesn't.
  3. Only a full restart of the VM seems to fix this. I have a free ChatGPT account so I will have to wait until this afternoon to continue solving in a way that it can automatically recover. At least I will try to improve the mechanical connection.
18

hmmm, yeah that was “ Details “ which could have been in your Topic description, when asking for advice for “ Connectivity Issues “

Also, in future try to use the best basic simple network “standard” setup for you Network
Changing back and forth with i.e “ Reserved vs Static “ is not something you should do as a troubleshooting purpose, as it just makes yet another “point of failure” ( Confusing the Router )
Even Changing Router, i.e to OpenWRT, is something that “initially” will affect your Network.
And Using static IP ( Setting fixed IP on a Device ! , is Configure Your Network Settings On That Device, to “ Live “ without the Services your Router Provides for the rest of your Devices) is only something you should use, If you really know your Network topology And your Router, And The Specific Server , build-in services and capabilities.

Yes, i know i might sound “harsh” , But seeing

Just “ woke me up “ ( Literately ) I just woke up :wink: , and this was kind of my first “ News to the world “ :laughing:

PS: After checking the Weather-Forecast

19

That'll do it lol

I'm glad you found it! That is prob one thing AI is better at because none of us respond as quickly so you can't really do real time troubleshooting.

20

Well, I had no idea this was the cause of the problem until I found the connect/reconnect event in the host laptop log. The reconnecting happens within a second so it's easy to miss.
Thanks for your advice, I am still learning about networking, it's quite a complex subject I find. I changed router because of a different problem, which also led to my HA instance not connecting well. My previous (provider-provided) router seemed to have issues connecting ethernet and wireless devices to each other, but it was not possible to go deep enough into the settings to fix this. The static IP seemed a reasonable method to check if DHCP had anything to do with it, which it didn't, so I changed it back to the default method. Reserving the IP address is necessary though, since some devices need to know the IP of my HA instance. There is also the .local address but I have had issues in the past where that didn't work well.

Anyway, the cause is found. There is a hardware solution of course, but apart from getting a cable that holds better, will also look for a software solution that detects this happening and can solve it without shutting down my HA instance. Or I might just bite the bullet and learn how to set up Proxmox, if that is more resilient after disconnect/connect cycles.