I recently bought the new Flint 2 and thought it would be a great way to learn some networking that I've always avoided. One of the things I thought might be something worth trying is setting up another wifi SSID that is exposed to tailscale for some of my IOT devices.
To test the configuration, I:
Set up a new SSID, EZ1-TS. Configured exactly like the default SSID
Set up a new interface, LAN_TS. Set up exactly like the default interface, LAN, but without any devices connected to it. Also the DHCP assigns IP's on 192.168.7.X instead of 192.168.8.X
Attached the firewall policies for the default lan to the lan_ts interface for testing purposes.
The issue is that when I'm on my normal SSID, I get 500 Mbps, and when I'm on EZ1-TS, I get less than 1 Mbps. I switched the EZ1-TS wireless SSID to the default lan interface I get the full speed, so I believe the issue is with the lan_ts interface. However, I cant figure out whats different from the other lan interface as I've copied all the settings but the device setting. I've even ssh'd in and copied the config from the lan interface to the lan_ts interface (just removing the device section) .If I attach a device to the lan_ts interface, it breaks the interface and I can't connect to the SSID and get an IP.
Do you have any suggestions on what I need to do to get it to work at reasonable speeds?
I haven't set up tailscale on the router yet. I'm fairly confident on how to set up tailscale for the subnet, but wanted to make sure accessing the internet is fine before I set up tailscale. It currently doesn't have the tailscale plugin installed.
The issue seems to be the LAN_TS interface (or bridge) configurations. If it copies as the default LAN, it should same speed with the LAN. To try to move one LAN (like lan4) port from the br-lan to the LAN_TS, and test the cable LAN speed.
I've got the same issue here - although I didn't copy from the default Lan bridge, I simply created a new interface using a subnet of 172.16.x.x or 10.0.x.x and didn't attach any devices at all.
For firewall zones, I used the preconfigured guest zone to allow outbound wan access to the internet.
Each time I attempt to change an SSID AP to use new said interface, I get download speeds of < 1mbps.
Strangely enough, this doesn't occur if I use the pre-generated br-guest interface and am able to get full speeds (400-500 symmetric)
As you can see I've attached the new network interface to the 5g SSID/AP - and the speeds are < 1mbps.
This is the case if I attach it to any of them - 2.4ghz or the 5ghz.
Bear in mind I've repeated this process twice and factory reset after each.
Steps as follows:
Login to router for first time and set credentials and SSID passphrases
Open Luci and create new interface using different private subnet range (192.168.20.x in this example), enable DHCP and create new firewall rule name
Go to firewalls and setup the new rules to reject input, allow output and reject forward (within same zone), and setup forward zone from new firewall zones to WAN.
Go to wireless and attach SSID/AP to the new network interface just created.
Use phone/laptop to connect to WiFi AP and check IP has been assigned in correct range (correct), check outbound internet access (correct but slow), perform speed test resulting in < 1mbps download and upload compared with ~500/500 when attached to default LAN interface.
I've tried both setting checkbox for default gateway and without in the new interface but no effect (as expected).
@bruce I wonder could it be a conflict with the gl qos?
it's just a observation i notice with multiple people, they either use sqm or they use a different interface and report similar issues, i could be mistaken.
are there any ethernet cables involved which are
damaged or have not properly being terminated and showing the coring? (Fyi, i have been in new bought homes before and people got told these bad cables where ok, which is bad 1/2 of the cable should be in the jack (I exaggerate a little) and only a small piece of the coring should contact the points, if the coring is outside the jack it's wrong).
Could it be that there are devices to far away taking away priority on the wireless link?
Could it be you have some areas in your network which is the cause of severe bufferbloat?, like if someone downloads half the devices dont work?
is the router powered differently than the normal power coord?, or is there a usb hub attached and perhaps a cable is close to the antennas?
^ i do know that flint 2 are extremely vulnerable to close things on the antennas.
Sure thing, happy to help as best I can! Thanks for the reply.
Topology: a star - fibre to the home through ONT which connects to the wan on the router. 4x devices connected via direct ethernet (things like raspberry pi, Nas etc.) and all other devices via WiFi (desktop, laptop, TV etc.). Small home network.
No usb connections involved.
Wan connection is wired, most clients connect over WiFi (not many clients in total, maybe 10 or so).
No ethernet cables that are damaged - I've simply reconnected them over from my current ISP provided router into the flint 2. Haven't tested ethernet speeds yet, was mostly trying to get the WiFi working with the new interface(s) first.
Distance isn't an issue, all devices are in close range. Nor is bufferbloat - haven't experienced any of this before. Additionally when doing the testing and getting sub 1mbps I only connected 2 clients via WiFi - the laptop to do the initial setup (connected via the SSID linked to LAN interface) and my phone which was used to test the new interface with the slow speeds (I additionally verified using other devices afterwards such as the laptop and desktop).
Powered directly using the power cable into mains - no usb power or hubs involved.
it shows two wireless phy interfaces on lan, but also one on your lan_ts interface, that seem not good.
can you try and check the following:
go to gl ui -> advanced settings -> luci -> interfaces -> tab devices tab -> edit br-lan and remove the rax2, you need the one being in use on the lan_ts interface if you hover over its icon you will see a name
^ if there isn't any rax1,rax2 inside the bridge that is good too wifi phys hook automaticly, by default openwrt it isn't there either, but i know from gl firmwares it often is there, that is not a misconfiguration perse but DSA design, however if a different interface uses it, it cannot use two interfaces on one device, you can view it as ports technically.
can you then navigate to luci -> network -> wireless and check if there are 2 network accidentally checked on the wireless access point?, this can been happened by a script.