Hello,

I have regular (2 times per hour) connection resets on the 'wan' interface of my OpenWRT 12.09 (r36088) router (D-Link DIR-615 D3):

netifd: Interface 'wan' has lost the connection

I have already tried to search for similar situations, like:

https://dev.openwrt.org/ticket/14144
https://dev.openwrt.org/ticket/11679
https://dev.openwrt.org/ticket/12114

However, in my case, there is no DHCP renew or any other messages before the resets:

Dec 24 15:35:53 OpenWrt daemon.info dnsmasq-dhcp[1604]: DHCPINFORM(br-lan) 192.168.xxx.yyy zz:zz:zz:zz:zz:zz
Dec 24 15:35:53 OpenWrt daemon.info dnsmasq-dhcp[1604]: DHCPACK(br-lan) 192.168..xxx.yyy zz:zz:zz:zz:zz:zz PCNAME
Dec 24 15:39:07 OpenWrt daemon.info dnsmasq-dhcp[1604]: DHCPINFORM(br-lan) 192.168.xxx.yyy zz:zz:zz:zz:zz:zz
Dec 24 15:39:07 OpenWrt daemon.info dnsmasq-dhcp[1604]: DHCPACK(br-lan) 192.168.xxx.yyy zz:zz:zz:zz:zz:zz PCNAME
Dec 24 15:40:14 OpenWrt daemon.info hostapd: wlan0: STA zz:zz:zz:zz:zz:zz WPA: group key handshake completed (RSN)
Dec 24 15:40:14 OpenWrt daemon.info hostapd: wlan0: STA yy:yy:yy:yy:yy:yy WPA: group key handshake completed (RSN)
Dec 24 15:40:14 OpenWrt daemon.info hostapd: wlan0: STA xx:xx:xx:xx:xx:xx WPA: group key handshake completed (RSN)
Dec 24 15:43:43 OpenWrt daemon.notice netifd: Interface 'wan' has lost the connection
Dec 24 15:43:44 OpenWrt user.info firewall: removing wan (eth0.2) from zone wan
Dec 24 15:43:45 OpenWrt user.notice root: stopping ntpclient

I have also called my ISP and asked them to check the modem. They said that it's fine and suggested that I should connect the laptop directly to the modem, to be sure that the problem is in the router. After I did it and tested the connection for the whole night using 'ping -t' output redirected to file, I found only 3 'Request timed out' messages for ~10 hours. As on the router I have connection resets about every 30 minutes, I think that the problem is definitely within the router.

Do you have any ideas how to fix/debug it?

Thank you in advance

Does the following command cause your connection to drop immediately, instead of 2 hours?

kill -SIGUSR1 `pidof udhcpc`

Can you post the output of "route" before and after?

That's your problem.  I had two of those; both exhibited the same behavior within weeks of being put in service. Bad hardware, bad design, bad manufacturing, I don't know but both units are long recycled....