Flint 2 (GL-MT6000 ) - bug reports - collective thread

i just registered & downloaded to try it out.

testing with over one dozen IOT devices on guest wifi 2.4Ghz

Also — download proceeded via HTTP, was prompted with a warning stating that download file could be dangerous due to tampering.
Not super important but nice to know that it’s safe in transit…

EDIT:
V4.5.5- Dec 12 ,2023
Bugfixes

Update WiFi firmware.
Fixed hardware acceleration enabled, unable to connect to the network after lower level wds.

https://fw.gl-inet.com/firmware/mt6000/tmp/mt6000-1226.bin

Should also work.

right, and the speed in 2.4 with 40hz exceeded 100mb? and the latency dropped?

You have to test more because the problem of dropping the WiFi connection and not being able to reconnect took a while, like when I went to sleep and when I woke up I could no longer connect just by restarting, and the last version 4.5.4 was presenting kernel problems that caused the router restarts, did you solve that too?

remembering that the best and most stable version possible is 4.5.2 which I have everything activated, adguard, wireguard, docker, sqm, wifi, cable connecting devices and I have no disconnection problems BUT still with the problem of 2.4 at 40hz not passing the 100mb and have enormous latency.

I had commented in other posts that the visible difference between 4.5.2 and the others was that WED was activated, perhaps this could also be related to the bugs they contained.

Yeah. I’ve been skimming this thread off and on for a few days.
My current workaround is planned restart every morning at 4AM.
It is not GREAT but provided more stability.

My situation is also a little unique b/c my ISP (in order to activate a deal) - requires me to use their router. So my current set up is:
MODEM > ISP ROUTER > FLINT2/MT6000 > Computer

Speed test on phone, within 1 m of router, yields the following:
34 Mbps , which is bad.

Wired speed test from computer shows:
340 Mbps, which is above average, but I sometimes get this.

restarting every day is not a viable solution for a router with such high power, it doesn’t make sense, the safest option is to restart it once a week if you want, the right thing to do is to work straight away without having to restart and if it starts to get slow you restart manually.

about the wifi apparently in 2.4 it didn’t solve it… it still holds up poorly.

Can you share the log with us to see if there are any errors?

MT6000 2 NEW ISSUES - too high CPU load / both MT6000 stuck offline in GoodCloud

1 - Too high CPU load in Administration Panel chart

First a positive remark: in general my two MT6000 units are performing well, one with around 70 devices and one with up to 80, without any issues for any of the connected devices. No further reboots after changing to v4.5.4 and the Wifi issue I mentioned before (see my previous post) seems to be not happening with the second MT6000 (stable 5 GHz Wifi speed of about 500-600 MBbps at a few meters distance). Since I am using a high quantity of connected devices I regularly check the CPU avarage load chart (under System - Overview) and it always stays below 0.5 for both MT6000 units:

Chart

But a few days ago, while I was connected through WireGuard to one of my two MT6000 and checking for some other parameters, I suddenly noticed that the CPU average load chart showed a very high value: between 5 and 6!!! I did not notice any performance issues at that moment, while the CPU load remained stable at around 5.5 (which should never be possible with 4 Cores),so I forced a reboot and after that the CPU load was back again below 0.5, with the same device load as before, and I have not seen such a high value afterwards.

A few questions:

  • how is it possible to see a CPU average load of more than 4? Does it mean that the load was really more than 4 or could it be that the chart was giving wrong values?

  • is there any explanation to see suddenly such a high CPU load whereas at all other times the load remains under 0.5, with no change of work load or number of connected devices?

2 - Both MT6000 stuck offline in GoodCloud, no way to connect them again

When I tried to check the CPU load through the GL.iNet iOS app (latest version 2.4.4), to which I had previously connected my both MT6000 without any issues, I found that the MT6000 was suddenly shown offline on GoodCloud, whereas I was working on it through Wireguard and it was not offline at all. A reboot did not help and I tried several ways to get the router back to normal status in GoodCloud but nothing worked. And while disconnecting and attempting to reconnect my second MT6000, it also got stuck offline.

So for now I cannot use GoodCloud anymore for my 2 MT6000, whereas for my other GL.iNet routers I have no issues using GoodCloud. See below pictures for the offline signs next to my 2 MT6000 and the wrong message I get in GoodCloud that the router is offline:

On the GoodCloud page inside the Administarion Panel it is shown that the ‘device security key verification has failed’, and there is a link to ‘Rebind the Device’ (in blue). However, clicking on this link only opens a Whitelabel Error Page without any further results (see here below).

In the Goodcloud App there is a Re-Activate option but this only results in a red ‘Permission error’ message (see below). There is also a ‘Bind to Cloud’ switch, but moving this on and off does not help either. Finally it is written to ‘clear the GoodCloud Token or go to FAQ’ but nowhere I could find any further information on this…


As a result, my two final questions are:

  • how can I get my 2 MT6000 routers back into GoodCloud?

  • could these bugs be corrected so that in the future it will be possible to get our routers back in GoodCloud in a more convenient way when they have been disconnected?

Linux is way more complicated than just having 100%
See Load (computing) - Wikipedia

I would recommend you to create 2 different new topics for your problems. This collective thread is already too long.

@lincoln @alex_zheng @alzhao @Dipin @JerryZhao

If you guys are forking the mt76 driver to include changes such as this one could you please upload your fork to GitHub and/or send pull requests to the main codebase? That way the OpenWrt firmware can benefit from the same fixes for our device.

And while I’ve got your attention, could you please update your gl-sdk4-ddns package so that it doesn’t include modified scripts from ddns-scripts? Since they use the GPL v2 license, which is a problem since you’re not sharing the changes and the old scripts aren’t compatible with add-on packages.

4 Likes

@lincoln and @alex_zheng

Lets try this firmware version.

Using openwrt image, i got freezing wifi too.

About the Speed of 5Ghz i got 1Gbit or 1.2Gbit

My Flint on version 4.5.4 hasn’t missed a beat for the last week. Then I enabled my TV with WiFi and switch on my Zidoo Media Player connected by Ethernet to watch a movie. After a minute it freezes. Rush over to my RPi4 thinking it’s frozen and see the flashing LED on the Flint2.

It wouldn’t re-establish and everything down. I have about 8 devices on WiFi.

Tried again after powering off and same thing. Turn off TV and Zidoo and try the Google TV on another TV, no problems.

Tried disconnecting Zidoo LAN cable and try WiFi, same thing occurs. Never had any issues before.

Still trying to work out what has caused this to cause a panic on the Flint2. I’ve got logging enabled to the /tmp path, but when I look, I can’t find the system.log file using WinSCP. I fear as I had to power off the Flint, it’s lost the log file.

UPDATE

it was the Panasonic TV wifi that was killing it. I couldn’t delete or disable the setting for the 5G connection so I had to update my wifi password to test the theory, flint was fine, then set the TV to the new password and the revert the password on the flint back to the previous one. I do hope the firmware fix mentioned yesterday fixes this. I can’t see it in the beta download section yet.

I tried updating it. It does say 4.5.5 and that makes sense I guess. I also installed it without keeping my settings. However, when I try to reinstall luci-app-sqm it shows errors now.

image

I hope it can get fixed ASAP since I need SQM on my current setup. Thanks! @lincoln

Replace the software source:

sed -i ‘s//kmod-4.5///kmod-4.5.5//’ /etc/opkg/distfeeds.conf
opkg update

1 Like

I will add my own stupid little bug.
From the GL interface, you cannot define port ranges in the port forwarding page, only single ports
In Luci there is not issue.

That’s not a bug but a feature request. May you add it here? List of current feature requests 2023

It works now thanks! For those who wants to also try you can use this command instead:

sed -i 's/kmod-4\.5/kmod-4.5.5/g' /etc/opkg/distfeeds.conf


Please try the activation button on the cloud.

[quote=“ptselios, post:237, topic:35608”]
From the GL interface, you cannot define port ranges in the port forwarding page, only single ports
In Luci there is not issue.
[/quote] Port range forwarding will be supported in Firmware 4.6.

2 Likes

This helped me to get my both MT6000 back online, many thanks for this advice. The conlusion is that you should review and fix the re-activate button in the iOS app which did not have any effect (as described in my post) as well as the 'Rebind the Device’ option in the Administration Panel which currenly results in opening a Whitelabel Error Page.

Btw, update on my 5G Wi-Fi problem. MT6000 1226 firmware aka version “4.5.5” did improved my 5G Wi-Fi bandwidth. It’s not as high as what my physical hardware can handle but I guess theres just some problems with my overall test setup. From 60 Mbps to this.

1 Like

The issue will be fixed in the cloud server update within this week.

1 Like