I have two and both behaves similar. No issue for me with WAN port on unit 2. Maybe both are HW1.0, but I don't remember. Number 1 is using LAN2 for unknown reasons.
Just dmesg either from the serial console, or ssh (or Luci system logs) after connecting on one of the 1Gbps port (since here it is for a boot where LAN2 doesn't work).
Also, if you can dump u-boot (/dev/mtd0, named vendor or bootloader), either using ssh or in Luci (System => Backup / flash firmware), maybe I can try to look at the Ethernet init code.
I just plugged in one of the ports to be honest, but I can change to LAN2 and see if I get some issues. I'm gonna do an upgrade to 6.18 on number 2 soon, so I'll make sure LAN2 is used.
Finally got a couple logs and some "save mtdblock"/dumps for you. A cold boot which happened to have no 10Gbe working at all, then a reboot where 10Gbe Lan2 works, but no WAN.
Both boots are strange, with the Link is Up being very late.
On the cold boot, LAN2 is at 1Gbps, then 100Mbps, then 1Gbps.
And even on the reboot, you said the 10Gbe LAN2 works, but the trace tells 1Gbps.
This doesn't look like any of the other cases, which OpenWrt build did you use ?
I guess I should clarify the laptop I'm testing most on is just 1gbe, but the WAN link that I test is directly to another W1700k on its LAN2 port. It shows as 10gbe properly if I swap ports so the link speed isnt the problem for me, its usually the "no RX" and not giving out any IP problem. For example on the cold boot I had to switch to 1gbe to get luci. Then on the reboots I can switch to Lan2.
Edit - I just noticed this build was actually still "realtekfix3" since I forgot to go back to the "test" build I have on the others. Was about to re-log it all for you, but the behavior has been consistent regardless of firmware for me.
I apologize, the WAN port was actually on the 1gbe port. I mustve switched it around when I thought I was done with this for a bit. But again, the "link speed" was never the problem.
realtekfix3 seems to help for the "no RX" issue which happens often on AN7581 rev E2 (ie HW1.1 it seems).
It doesn't change anything on older SoC, and I don't see any failure trace about the calibration, so it could be another issue.
Yeah it sounded like it might've helped when I saw it discussed earlier which is why I tried it. But then later on when I found that devmem command I figured it wouldn't affect me because of that?
With how consistent my problems have been, and I honestly cant even remember when the WAN port worked last. Maybe mine is actually a hardware issue? I kind of wonder how many are DOA or defective out of the box for when Quantum Fiber actually deploys them.
Lan 2 works pretty reliably if I leave it unplugged on boot and plug in later. It pretty much acts like the "reboot" log I sent earlier where I left 1gbe plugged in. But almost every time I've left unplugged it'll give me an ip and it seems to function as normal on replug on LAN 2.
Based on that I figured I could rebind WAN to that port, but yeah the WAN/LAN1 10Gbe port remains the same "no RX/IP".
Edit -
As im thinking more about it, its the cold boots or hard resets that seem to mess with it and cause having to unplug/plug ethernet to get it back. Reboots afterwards seem to be ok.
Here you are using LAN2 as 1Gbps, so maybe it makes a difference.
Are the other w1700k which work better also with LAN2 at 1Gbps ?
You might still try in the vendor u-boot :
setenv serdes_ethernet
saveenv
This will skip the serdes init (ETH VER = ETH.2.2.1-R ... USXGMII_10G exit).
It will help if there is some kind of conflict between what u-boot set and Linux drivers, but I don't know why the same code would cause different issues depending on the device and/or environment.
The default value 11 initializes SERDES_ETHERLAN (so probably LAN2) in USXGMII mode.
But later the code seems to only use the switch (so the 1Gbps ports) to send packets, since fw_port=0
By hard reset I mean if i unplug/plug power back in to restart. If i reboot through command or luci I can pretty reliably get Lan 2 back up.
The other W1700k's I have two connected to each other, main AP has 10gbe ONT plugged into WAN and its 10GB lan2 goes to the other W1700k/dumb AP on its WAN, which then goes to my main PC usually on the LAN 2 port. The other 1gbe ports are used for switches at the moment for other devices. Basically goes ONT > WAN W1700k1 from LAN 2> WAN W1700k2 from LAN2> Main PC, switches on 1gbe(Lan3,4).
All 10gbe ports report 10gbps properly on the other units, and im 90% sure the problem one does as well from what I remember on initial experience (besides its WAN port which again I genuinely cant remember if I ever had working)
For me, unplug/plug the power is a cold boot.
So I don't understand what you meant with "cold boots or hard resets" (written as two different things).
Logs may be useful when comparing similar configurations and environments.
For example, you can first take logs of a cold boot of w1700k2 connected as you described.
And then configure w1700k3 like w1700k2, and either replace it or chain it behind (since it should be similar, it would have a w1700k on its WAN, and the PC on its LAN2).
Then w1700k2 / w1700k3 should have very similar boot logs, except for random timing variations (which could create bugs, but it wouldn't be enough be explain why one would work reliably while the other would fail often).
Yeah sorry, maybe I should've clarified on the cold boot/hard reset (I mentioned earlier, it was getting late) really mean the same thing in that use case. Can almost always get Lan2 to fail if I unplug/plug in power to reset, requiring a bunch of unplug/plug switching just to be able to access luci/reboot. Doing a reboot through command or luci I can usually always get it to work. But back to a full unplug/plug in power and it usually fails again.
Ill try to get some logs of that configuration for you. This third one really was just here as a backup and to test new firmware configurations before implementing into my main loop. I have tested it as replacement for both my "W1700K1", and "W1700K2" in my setup before but obviously unsuccessfully as 10gbe ports were so unreliable. I'll have to wait for some downtime in usage before I can test though.
I posted a similar message in the community builds thread earlier:
Is anyone using one of the snapshot build with a Samsung Galaxy S26 Ultra?
I'm still on a pre UBI fan boy build and my S26U chokes after a few minutes, send and receive speeds plummet and it often swaps to mobile data. The S25U had no problems at all. I don't have the same problem with the S26U on the WiFi at work. I'll test with my old Belkin RT3200 later today to see if that works OK.
yep im using S26 ultra with fanboy's test version. I have the same issue where my speedtest went down to about 1/10 of the full line speed after sometime. It can be fixed by reconnecting the phone to the wifi
I’m sorry you’re seeing the same issue but I’m glad it’s not just me.
I created an additional 802.11ac 5Ghz network a couple of hours ago and the wifi has been stable so far. 5Ghz, AC, 80Mhz, channel 36, WPA2. Hopefully that will keep working until Samsung release a firmware fix.