WAN speed issues with the AX86U

[OFark]

Hoping someone with more experience than I can help.

The History:
Some time ago I upgraded to the aforementioned Asus AX86U with Merlins FW, to solve some wireless issues I was having with an older router. It's been great until Virgin upgraded my internet from 350Mb to 1Gb, accompanied by a new Router. Initially, everything was fine, but then after a few days my connection deteriorated (more on that later), I restarted everything and it was all back to normal. For a few days. Then connection issues, restart the router, back to fine. A few days later, repeat. And again, only this time I restarted my PC, no change, restarted Virgin's Router, no change, restarted the Asus AX86U, back to fine again.

The Specs:
The Virgin Router is their latest Super Hub 4, in Modem mode.
The Router is the Asus AX86U FW: 386.3.2
Everything in consideration is wired. CAT5e
Speedtest on the router won't pass 700Mbps when the PC will record 950Mbps (both when everything is ok)
50Mbps up on both.

The Connection Issue:
Speedtest down drops to 80Mbps (Both router and multiple PC's) whilst the upload remains at 50Mbps
Apparently, my voice sounds digitally choppy to other gamers (Xbox/PS5) and they occasionally cut out for me, but never distorted.
In my Teams meetings, my camera looks really bad quality, like it's making up for poor lighting, voices often go into what I would describe sounds like slow motion.
In some games, I get really bad rubber-banding.

The Router logs don't say anything that jumps out at me, but I'm not sure what I'm looking for. I do get: nf_conntrack: expectation table full, but following that around in this forum, I seem to be the only one experiencing slow down.

Worth noting:
I am using Diversion
I am maintaining a VPN tunnel out to Nord, for certain clients, but not this PC.
I am using a VPN Server, on the router, for my phone to connect back in again.
Hardware or software reset fixes the problem for the same length of time.
CPU and RAM are not maxed out.

What I need help with:
I am inclined to believe the problem is with the router, simply because resetting it seems to cure the problem for a few days. But I appreciate that the new Virgin router has triggered the issue, and I am open to suggestions, but I need proof if I am going back to Virgin with an issue. Otherwise, I am looking for something to try on the router, or ways to diagnose the problem, things to look for in the log, or settings etc.

Thanks in advance.

 

[ColinTaylor]

Speedtest on the router won't pass 700Mbps when the PC will record 950Mbps (both when everything is ok)

This sounds about right. The router's built-in speed test becomes CPU bound on one core at about 700Mbps.

If the Hub 4 is anything like the Hub 3.0 you can go to http://192.168.100.1/ and check its logs. Check the Network Log and Downstream Post RS Errors.

 

[L&LD]

An RT-AX86U on a 1Gbps symmetrical Fibre connection will show ~940Mbps up/down when no scripts, features, or options (such as VPN servers/clients) are running. This is with the latest 386.4_alpha2-g952c6bdecc firmware available.

When you perform a hardware reset, are you using the method below?

[Wireless] ASUS router Hard Factory Reset | Official Support | ASUS Global

In addition to formatting the USB drive in a PC to NTFS before inserting it back into the freshly reset router? And, not using a saved backup config file to secure and configure the router too?

At this point, I would be considering flashing the Alpha firmware I note above (at least for testing). Note that all the available Alpha's have been very stable for me. This latest one has been running for 20 days and 19 hours so far (since it was released).

Turn off the Hub 4 until you've completed the steps below (and keep it off for at least half an hour, an hour or more is ideal, even if the steps below don't require that much time).

Next, I would follow the link above to fully reset the router. Take this opportunity to format the USB drive containing Entware and your scripts on a PC w/NTFS format.

Now, without connecting the router to the internet, configure as much as possible on the router. But minimally, as described in the link below.

Fully Reset / Best Practice Setup / More


With the Hub 4 powered on and left to settle down (at least 5 minutes after it boots up), connect the RT-AX86U and power it up.

Now, insert the USB drive in the router and slowly and methodically format the drive to Ext4 w/journaling, create a 2GB swap file, and test the network for performance and stability.

If the router/network remains stable and performs as expected, add one change (feature/option, script) to your network and test extensively again, including a reboot after an hour or more after making the last change.

Keep good notes and you'll soon find where (or why) the issue arises for your network topology.

The symptoms you're describing now are not normal in any Asus/RMerlin installation I've seen in the last year.

HTH

 

[OFark]

This sounds about right. The router's built-in speed test becomes CPU bound on one core at about 700Mbps.

If the Hub 4 is anything like the Hub 3.0 you can go to http://192.168.100.1/ and check its logs. Check the Network Log and Downstream Post RS Errors.

I figured the router wasn't fast enough to do the speed check, thanks for confirming.

I've had a look at the logs and well I don't understand any of it, but there hasn't been an entry in 3 days, and that was DHCP Renew. There are a few "No Ranging Response received - T3 time-out" messages that occur around 3 AM but again that was a few days ago, and this issue has occurred again this morning.

 

[OFark]

L&LD, Thank you for the message, you've helped me before and suggested this router. To be honest I was a little afraid of an answer like this, as it says to me this router is so flakey you have to reset the whole thing back to factory settings to get it to work properly again. I mean the last Asus router I had you had me hardware reset it twice because sometimes the first time isn't enough. Maybe it's just Asus routers, maybe they are really flakey. Maybe they are not for me. I don't have the time to be powering down anything for an hour. Working from home, having a family that use the internet I just don't have time to be enabling one thing and waiting an hour.

I appreciate the help, I really do, but isn't there something I can do/check/try before I turn the router into a full-time-unpaid-support-job?

 

[L&LD]

I hope others can give you some quick-fix suggestions. I just know by your description of the problem that the router isn't in a good/known state to be exhibiting this behavior.

What I've described above is what I've done on every single Asus/RMerlin powered router I've deployed for many years now. For customers, I would repeat that for them (often free of charge) when major updates became available, but for at least the past year, I take a wait and see approach (to see if a full reset is needed). Out of many, many installations, I have only had to do a handful 'from scratch' or the popular term on the forums here, go 'nuclear' option.

I can understand that time is limited to try such things, but for whatever reason, some network environments need it. Ultimately, it is your choice.

But from a hardware point of view, I am confident it isn't merely the hardware or the firmware that is at fault here. The Asus/RMerlin combination isn't flakey, IME, rather the contrary, when it is set up properly as I've outlined above.

(FYI, I've not reset my routers since setting them up almost a year ago (and it is roughly 2 years since I've routinely been resetting my own routers with new firmware), through beta, release, and now back to Alpha firmware, and they're still not exhibiting any anomalies as stated here, in this thread. But they were nuclear reset a few times when they were new and after I installed the RMerlin Beta firmware back then).

What I can offer for you to check verify is to start at the basics. Are the cables used in the network in perfect condition? Does simply powering down the network for at least 30 minutes, disconnect all data and power cables from all equipment, then powering the network up in order (modem/ONT, router, switches, NAS, printers/scanners, client devices) help? Does flipping the Ethernet cables (end to end) help keep a solid connection/speed? Does adding a cheap 5 or 8 port, unmanaged switch, between the routers help?

Do you have power issues at your location? Is all the networking equipment on quality UPS'? Have you had your ISP send a tech out to check the quality of the equipment on their end?

I hope others have more directly applicable (possible) fixes for you.

 

[OFark]

So I've found something interesting in the log when it happens:
First:

Dec 11 11:59:08 kernel: eth5 (Ext switch port: 7) (Logical Port: 15) (phyId: 1e) Link DOWN.

Then some eth6(wifi ?) auth - deauths
Then:

Dec 11 11:59:32 kernel: eth5 (Ext switch port: 7) (Logical Port: 15) (phyId: 1e) Link Up at 100 mbps full duplex
Dec 11 11:59:37 kernel: FPM Pool 1: invalid token 0x206d0000 freed

I am assuming eth5 is the WAN port to my router/modem? Looks like something is dropping the gigabit connection. Nothing on the router/modem's logs.

 

[ColinTaylor]

eth5 is the 2.5Gb WAN port. I suggest you don't use that but instead use the normal 1Gb WAN port.

 

[OFark]

eth5 is the 2.5Gb WAN port. I suggest you don't use that but instead use the normal 1Gb WAN port.

I have the WAN into the normal WAN port (with the old IE logo) and the Router to the network on the 2.5, I'll try the LAN on PORT 1

 

[OFark]

eth5 is the 2.5Gb WAN port. I suggest you don't use that but instead use the normal 1Gb WAN port.

Isn't that 2.5Gb a dedicated LAN port, not WAN?

 

[ColinTaylor]

Isn't that 2.5Gb a dedicated LAN port, not WAN?

It can be assigned as either the WAN port or another LAN port. As you said you thought it was your WAN port I assumed that was the situation. In any case, whatever it is that's plugged into it is wasting its potential if it's operating at 100Mbps instead of 2.5Mbps.

 

[OFark]

It can be assigned as either the WAN port or another LAN port. As you said you thought it was your WAN port I assumed that was the situation. In any case, whatever it is that's plugged into it is wasting its potential if it's operating at 100Mbps instead of 2.5Mbps.

I ASSUMED it was my WAN port because I'm with Virgin Media. But looking at it, it was a LAN port 5 that was disconnecting. Thanks for that pointer, Internet has been stable all day.

 

[OFark]

Looks like that fixed it. Thanks.

 

[ColinTaylor]

Looks like that fixed it. Thanks.

What was it that you changed in the end?

 

[OFark]

Just moved the LAN from the 2.5G to one of the standard 1G ports. I have no idea why the 2.5G couldn't maintain a 1G connection, but there it is.

 

[L&LD]

It seems like your cable and/or ends are not compatible with the router's 2.5GbE Port.

Can't remember if you have tried with new Ethernet cable.

 

[OFark]

It seems like your cable and/or ends are not compatible with the router's 2.5GbE Port.

Can't remember if you have tried with new Ethernet cable.

It was on my list of things to try. Problem is the house is wired by my father and I. I used to be a network installation engineer, my dad was an electrician. We ran Ethernet to a central location for the router from the place where the internet comes in. For wireless coverage. So between the router and the modem is 3 cables, a patch panel and a wall plate. I used cat 5e. But there's a chance a pair was untwisted a little to much.