I have decided to try 24.10.1 on my edgerouter 4. I have managed to get an image loaded however as soon as I reconfigure wan port lan0 to be pppoe the lan bridge interface and port lan1 that I was physically connected to looses connectivity from the browser to Luci. I've done the failsafe to start over but this keeps happening.
Console is clear of errors right up until save the pppoe on the wan. Then this error appears for port 2 which lines up to lan1 in the bridge. If I move cable to another bridge port error moves there.
Now just to be sure I swapped out cable. Only other thing I thought is lan1 connected to 2.5G switch which autoneg down to 1G as you would expect. If that was a problem surely I would have these errors from start and never even reach the gui. Soon as I configure pppoe,creds and mtu as 1480 save and apply it strikes.
The lan1 port? No because that’s the got the management/gateway ip I’m trying to connect to. I don’t understand why that would or should be necessary or desirable.
We can probably figure this out by looking at the default state of the config and then recommending changes to do directly on that file.
Please connect to your OpenWrt device using ssh and copy the output of the following commands and post it here using the "Preformatted text </> " button:
Remember to redact passwords, MAC addresses and any public IP addresses you may have:
Do you mean the config as I have configured it (ssh connection to died same time as gui but could use console) or firstboot to restore then I would be able to ssh? The latter config being clean.
Your config looks OK, under normal conditions you use lan0 as ppp backend, which should not impact br-lan memb€rs in any way.
Try PADO discovery before saving.
I understand lan0 is mapped as WAN and changing it to pppoe from dhcp should not kill my management connectivity on the bridged ports but it does. I looked up PADO and that seems to be related to my internet connection stability which seems like putting the cart before the horse. I might try ditching the bridge altogether.
Tried taking the lan1 interface out of the bridge group. As before no issues connecting initially but as soon as configured all my pppoe credentials it died. So I started with clean install again and saved after each step configuring then moving back to status page. So far so good. Then I changed mtu to 1480 on wan/lan0 and that's when I lose connectivity. Flipping MSS clamping or mtu_fix in the firewall settings doesn't help either. ISP requires this fixed mtu so I'm kinda stuck. changing mtu on wan interface should not present issue to connections terminating on green/lan/bridge. So is this a bug?
GoFibre in the UK and any of the smaller providers that need to sublet from openreach backbone. I live in rural area. This is the only fibre to the home provider.
I don't know know what self-KO means. It was on default of 1500 yes.
This is my wan config as it stands at the moment and captured from console port. The cisco cable really working hard this week
# From /etc/config/network
config interface 'wan'
option device 'lan0'
option proto 'pppoe'
option username 'xxx'
option password 'xxx
option mtu '1480'
option ipv6 'auto'
and the MSS clamp setting. I get the logic of this as there ISP gets the same size irrespective of customer equipment or it's own.
# /etc/config/firewall mtu_fix set to 0 by me since 1 wasn't working.
config zone
option name 'wan'
option input 'REJECT'
option output 'ACCEPT'
option forward 'REJECT'
option masq '1'
option mtu_fix '0'
list network 'wan'
MSS clamp patches (small) syn packets to fit future connection inside ppp tunnel mtu, contrary to large from start udp which get "frag-needed" icmp feedback from the very start.
Yeah I saw that post as well when I was preparing to switch over and deliberately asked for mtu size. I was told be gofibre that 1480 was the setting and it's been working great on my other edgerouter that's still running EdgeOS. Irrespective of ISP settings I should not lose management connectivity by changing the mtu on the WAN interface which. I assume that gofibre needs the headroom as there is futher encapsulation from all local areas before it's forwarded on.
((gofibre-local)->(gofibre-central))->(openreach)