I wonder if that might only be happening on 2.5G and not <1Gbps negotiations. I am rocking SNAPSHOT, r27243 and have not observed that either. But I do not have a 2.5Gbps link in WAN
I'm running build r27229-ebe7c5f1a3 from Monday and I believe this is the latest mt7986 firmware.
root@GL-MT6000-1:~# dmesg | grep -i firmware
[ 0.000000] psci: PSCIv1.1 detected in firmware.
[ 12.145560] mt798x-wmac 18000000.wifi: WM Firmware Version: ____000000, Build Time: 20240823160721
[ 12.230484] mt798x-wmac 18000000.wifi: WA Firmware Version: DEV_000000, Build Time: 20240823160840
root@GL-MT6000-1:~#
It is possible, as I do have a 2.5 Gbps connection to my ONT and am having the issue.
Ok didn't notice the latest fw go in: firmware: update mt7916/mt7981/mt7986 firmware to version 20240823. Was comparing to see if it made a difference on the regression.
I have a 1G link and am having this issue.
Like therpa says above, this happens with a 1 Gbit/s link, too.
Has anyone tested r27266-2c48cda28b?
kmod-mt76-core (6.6.47.2024.08.25~904ef52a-r1)
kmod-mt7986-firmware (6.6.47.2024.08.25~904ef52a-r1)
Removing the timing change, based on the idea from @daniel in the bug report, seems to help.
This seems to work for me with r27269-c6f04f718e:
- delete the no-op cosmetics patch 12: target/linux/mediatek/patches-6.6/733-12-net-phy-mediatek-Fix-alignment-in-callback-functions.patch
- modify the patch 10: Keep mtk.h stuff but remove new functionality: target/linux/mediatek/patches-6.6/733-10-net-phy-mediatek-Extend-1G-TX-RX-link-pulse-time.patch
Details:
https://github.com/openwrt/openwrt/issues/16281#issuecomment-2322095969
My test build is at:
https://www.dropbox.com/scl/fo/51htl4fzsx8jhg5u70mln/h?rlkey=nn1lb2yxsnw1eqkaamtop3qjj&dl=0
Regarding the WAN issue, if I understand it correctly it only affects WAN's DHCP, doesn't it? Because I'm using static on my WAN and no issue so far. I'm on r27229-ebe7c5f1a3.
Yeah, it is some kind of race condition in interface bring up or link status change, so that sometimes/often there is no trigger for DHCP client (udhcpc from busybox).
Q: Is anybody still unable to use current SNAPSHOTs? I tried installing r27283 yesterday and mesh did not connect reliably and the main router was unavailable for minutes at a time then ran slowly. PPPoE would connect only after some time but the routers themselves were so unstable that I didn't get as far as testing Internet connectivity.
23.05.4 is fine so I'm using this version for a few weeks now. The last usable SNAPSHOT for me was r27154.
Just curious.
Hi, yes for me last working snapshot is r27157, after that the router is not working at all.
I noticed that not only the WAN port is continuously disconnecting but also all the lan ports.
I have a voip phone and a workstation connected by cable and I see the Ethernet interface going up and down continuously on both devices, it is not only a DHCP problem.
Even if I assign a fixed ip to the workstation it is not possible to connect to the admin console on the router, you have to time the time slot in which the interface would be on, sometimes I'm able to login but then it disconnect again and I have to wait.
After around 30 minutes at times it stabilize but it is then very slow.
Something got wrong after commit 1591e2d4ccfd80da836017efb996b6a4ab59dba6 I can use snapshot till the commit before that, of the 16 of august.
Today my GL-MT6000 will arrive.
I will be switching from an RT3200.
Before I set everything up completely new, can I import the saved configuration from the RT3200 or are there areas from the backup that I can safely restore?
not recommended at all
If vanilla OpenWrt versions and installed packages set match, there're typically 2 places where a problem might occur:
/etc/config/wireless -> wifi-device -> path. In case of 23.05.4, set '2g' to 'platform/soc/18000000.wifi'
and 5g
to 'platform/soc/18000000.wifi+1'
.
/etc/config/network. Make sure your bare metal devices are set to this:
config device
option name 'lan1'
config device
option name 'lan2'
config device
option name 'lan3'
config device
option name 'lan4'
config device
option name 'lan5'
config device
option name 'eth1'
Next, check interface
s and other device
s, like bridge
s, and make sure they reference device names above.
If you end up with a bricked router, it's incredibly easy to recover to factory settings and configure wireless and network manually. Then scp other configs. I encourage you to to the firmware generator and build an image with packages you need already baked in. Saves you from installing packages if you need to recover.
Thank you for your answers.
I think I'll go down the route of creating everything manually.
This will also refresh my knowledge.
Welcome aboard. When flashing the sysupgrade just know there are some minor bugs (e.g. 16281, 16279) with snapshots at the moment. Once resolved it'll be a big improvement, but for now I recommend flashing the latest stable with a clean config unless you know what you're getting into.
It is always recommended to get a backup of your config as a guideline to configure your new router.
So, the device is set up and running. I have set everything up manually according to the old configuration. The only things I copied were my static IPs and the firewall rules.
I have read about the problem with the snaps, so I went to stable and am waiting to see how the router runs once the bugs have been fixed.