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

Technical Support for Routers
43

I use the standard AP Router mode.

It sounds for more more like this bug, which happens with the Mediatek WiFi of the BPI-R3:

also this comment: Bananapi BPI-R3 freezes randomly · Issue #13198 · openwrt/openwrt · GitHub

The stack / call trace is almost identical and the symptoms are same:

the following conditions must be met, so that there is a chance to reproduce it:
You must have 2 GHz and 5 GHz radios enabled
Clients must roam between 2 GHz and 5 GHz
It may be necessary to have more than one ESSID on both radios and have the same ESSIDs on both radios
It may happen less often if the wireless status is queried periodically e.g. by collectd
It may happen less often if legacy rates are enabled on the 2.4 GHz radio
It might happen only with specific clients (People mention Apple devices, but I think for me the problem was also triggered by an HP laptop with Intel Wi-Fi card running Linux)

There is already a patch for openwrt but since GL-Inet became closed source, I can’t compile it by my own.

@japa182 , which firmware version do you use?

2 Likes
44

Could be a issue at the moment the mediatek driver is so new that also upstream alot gets changed, to OpenWrt itself aswell.

Also there is also a hostapd flushing issue which can can also cause crashing with wireless bridges / multi psk.

But its good you post this, so GL-Inet can look into it :+1:

For now i use roaming too with psk2 but no crashes yet im on 4.5.3 (from december month).

1 Like
45

When you use DDNS from GL.iNet you have to change the config because the DDNS is tied to the device internal name, afaik.

46

4.5.2 and 4.5.3 same error

47

i have a usb3 hub with 2 6tb hard drives plugged in it
this worked for the longest time on my wrt3200 running openwrt
but the flint 2 doesnt seem to work with it…
i tried plugging one of the hdd’s into the usb port, without the hub, but it still not recognized…
maybe ill create a separate post about it
dont see any errors in logs
the 2 enabled ones are my existing disk uuids, i manually edited the fstab file to see if that would get it to work

image
image1920×1143 113 KB

seems weird that by default those other not-present ones show up

update: single drive did work (2nd attempt) - maybe the hub is incompatible
update 2: upon installation I accidently loosened the plug for the usb hub (so it wasnt getting power). once plugged in and getting power on its own, all drives showed up in openwrt

48

What is the system and kernel log showing when you connect the usb3?

Does the usb hub come with a power supply?

1 Like
49

Kernel crashes is a new issue. We’ll fix it in the next beta firmware.

1 Like
50

Sorry, I didn’t see ‘Public WiFi’ until a few moments and then could not reply to my reply as I just registered and had a 24 hour hold on my account :woozy_face:

51

It’s incredible that here it seems to have gotten worse, before I would wake up in the morning and not be able to connect, now until the afternoon I’m having the same problem of not being able to connect to the WiFi anymore because it says that the password is incorrect or there is a connection error… just one moment of inoperation that gives the same problem… I’ll send my log to see if anyone knows the real reason for this problem…
system e kernel log.zip (36.2 KB)

52

daemon.err stubby[3764]: Could not schedule query: None of the configured upstreams could be used to send queries on the specified transports

Something is odd with your DNS or the internet connection itself.

xt_FULLCONENAT: warning: failed to register a conntrack notifier. Disable active GC for mappings.

That’s weird. Never saw this error before.
What non-default settings do you have active on your router?

53

Does your WiFi is connected but not Internet?
Some new Android phones hard to keep WiFi connect and no Internet access, while can control LAN network.

54

@admon @slesar pppoe connection, automatic dns, use tailscale, and also the network share a 1tb hd, the rest of the settings are all standard, wpa2 on 2.4 and 5ghz, factory default wifi encryption I just changed the name and changed the region using openwrt to brazil which is my location, I also changed the default ssh port to another number and to connect I use a key generated by puttygen,

I’ve already seen this fullconenat error here, but oddly enough it presents the error with the fullconenat function activated or deactivated

@slesar I have a connection, so much so that I can connect via tailscale and restart the device without having to unplug the power cord, but it doesn’t work on any of my devices, whether Android or any other, or it shows a connection error or says that my password is incorrect and before being inoperative everything was working normally.

55

Hi,

I have another problem. I can’t create Wireguard profiles. I get this error:
Unknown error occurred. Please check the network environment or reboot the device.

Screenshot 2023-12-06 at 12.50.50
Screenshot 2023-12-06 at 12.50.501322×692 39.1 KB

HTTP response to the POST:
{“id”:15,“jsonrpc”:“2.0”,“error”:{“message”:“Internal error”,“code”:-32603}}

nginx error log:
2023/12/06 12:25:09 [error] 4741#0: *57587 upstream prematurely closed FastCGI stdout while reading response header from upstream, client: 192.168.0.197, server: , request: “POST /rpc HTTP/1.1”, subrequest: “/cgi-bin/glc”, upstream: “fastcgi://unix:/var/run/fcgiwrap.socket:”, host: “192.168.0.1”, referrer: “https://192.168.0.1/”

Anyone else having this issue?

56

Did you reboot the router and the error persists?

What firmware version are you running?

57

Hi

Yes, I’ve rebooted.
I’m running latest stable firmware (4.5.3 release4).
I have tried with Wireguard turned on and off.

58

Version 4.5.4 is available, try testing it.

Edit: beta version, only via the website.

59

It says “Firmware is up-to-date.” in my router.

60

It’s okay, I will install it manually from GL.iNet download center

61

Okay, I’m now running 4.5.4 and still have the same issue.

62

@Raboo Could you please create your own thread? It will be easier to follow up this issue.