Loosing Network Connection every few days

I have Freedombox installed on Banana Pro. Everything works fine for couple of days. Then the Freedombox cannot be reached from my home network. Banana Pro is connected to my netgear router by ehernet cable and gets assigned ipv4 address. After couple of days, I can see that router still shows banana pro is connected, but I can not ping it, or ssh into it. I cannot reach it by web browser. So I have to pull the power cable and insert again to restart banana pro. This has happened every time, so for now it is shut down. I would love to get it working reliabky. One way would be to reboot it nightly by batch job. But I would like to fix the problem, not patch it.

Any idea why this is happening?

Try turning on permanent log storage in the ‘configuration’ app in system section. Then examine the logs using terminal or Cockpit.

Another way is to see if you can connect it to a monitor to see if the machine is unresponsive or if it is just the network connectivity that is broken.

1 Like

A second DHCP server on your network can do this. After you get the logs set up use cockpit and filter for DHCP entries. If you find another device serving addresses you can treat the problem like this.

The link tells you how to tell freedombox to ignore the bad DHCP server. This would happen to me from time to time as you describe.

Thanks for the suggestions. I have connected Banana Pro by hdmi cable directly to monitor and keyboard. I have logged in. I am also going to get the logs set up next. Will wait for a couple of days and see if monitor still has connection, and try to ssh into it from another linux pc on the network. I will come here and report.

So in less than 12 hours, when I got up in the morning to check, the ssh connection to Freedombox was gone. I could see the ip address in the router web login though. Also, the monitor where I was logged in before retiring last night, was different output.

Looks like the system tried to reboot at night, and could not. So I unplugged and plugged back banana pro to reboot Freedombox. I do have log file on the sd card root partition in banana pro. I looked at the log, and it appears that when I booted in the morning by unplugging and replugging banana pro, the time was wrong, and during rebooting, it adjusted time, see these few lines from log. Note that correct time was 10:25 am, but during reboot it was set at 3:04 am.

Oct 12 00:03:04 debian fail2ban-server[422]: fail2ban.filter [422]: INFO maxLines: 1
Oct 12 10:25:44 debian systemd-resolved[238]: Clock change detected. Flushing caches.
Oct 12 10:25:44 debian systemd-timesyncd[239]: Contacted time server 192.48.105.15:123 (2.debian.pool.ntp.org).
Oct 12 10:25:44 debian systemd-timesyncd[239]: Initial clock synchronization to Sat 2024-10-12 10:25:44.612034 PDT.

I will appreciate any help. I do have the log file, so I can look at other things if you suggest.

Thanks.

If you have not done this yet, I would first try putting the microSD card in a reader in a PC to check the file systems.

Before inserting the microSD card, run sudo systemctl stop udisks2.service (assuming you have a GNU/Linux PC using systemd) on the PC so that it does not try mounting it automatically. If it is like on the Pioneer, there is a boot ext4 file system (use fsck.ext4 to check it) and a root btrfs file system (use btrfs check).

It is likely that like for many Single Board Computers, Banana Pro does not have an Real-Time Clock to keep track of time during reboots. The NTP running on FreedomBox is meant to correct this as soon as it can from the internet. So, ignore those time related inconsistencies.

The reboots are scheduled by FreedomBox when there is a security update that requires it. It seems that uboot is stopping mid-boot as if someone has pressed a key (to allow for debugging boot process and for custom boot options). Are you sure there was no keyboard activity at that time? Since rebooting the machine again worked, I guess this could be a one-time error.

You are correct. I remember rebooting on the web cockpit, and then going to monitor to check if the login prompt appeared. When the screen was blank, I had hit some keys to get attention. But as you pointed out it was u-boot booting, and my keystroke interrupted it.

I did run one command that may have helped, and for last two days, I am not loosing network connection. The command was:

hwclock --systohc

Since I ran this command, I have not seen time changes in the logs. I guess my cpu clock did not have correct time, and the command set it.

I am still watching. Let me see if the system looses network connection.

1 Like

I’d be surprised if the hwclock command fixed the issue. So, continue monitoring.