GL-XE3000 Puli terrible repeater performance

I started noticing this issue on 4.7.4 firmware, but it's also happening on the 4.8 beta.

On a fresh reboot, the repeater speeds are normal and about 90% of the original network's speed. Then after a few hours to a day, the speed drops to <1mbps. There are also constant dropped connections, and even the admin panel randomly shows timeouts or lack of internet connection.

The logs are full of the following line:

Wed May 14 15:15:28 2025 kern.err kernel: [  962.070125] 7981@C13L1,tx_free_v3_notify_handler() 3530: ContTxFailCntTotal = 2, ContTxFailCnt300ms = 2
Wed May 14 15:15:28 2025 kern.err kernel: [  962.079519] 7981@C13L1,tx_free_v3_notify_handler() 3533: token used by current wcid = 1, free_token_cnt = 2303
Wed May 14 15:15:29 2025 kern.err kernel: [  962.477707] 7981@C13L1,tx_free_v3_notify_handler() 3530: ContTxFailCntTotal = 3, ContTxFailCnt300ms = 1
Wed May 14 15:15:29 2025 kern.err kernel: [  962.487103] 7981@C13L1,tx_free_v3_notify_handler() 3533: token used by current wcid = 1, free_token_cnt = 2303
Wed May 14 15:15:30 2025 kern.err kernel: [  963.301153] 7981@C13L1,tx_free_v3_notify_handler() 3530: ContTxFailCntTotal = 4, ContTxFailCnt300ms = 1
Wed May 14 15:15:30 2025 kern.err kernel: [  963.310548] 7981@C13L1,tx_free_v3_notify_handler() 3533: token used by current wcid = 1, free_token_cnt = 2303
Wed May 14 15:19:40 2025 kern.err kernel: [ 1213.708704] 7981@C13L1,tx_free_v3_notify_handler() 3530: ContTxFailCntTotal = 1, ContTxFailCnt300ms = 1
Wed May 14 15:19:40 2025 kern.err kernel: [ 1213.718107] 7981@C13L1,tx_free_v3_notify_handler() 3533: token used by current wcid = 2, free_token_cnt = 2302

I noticed several older threads have this same issue:

And more.

My primary use of this router is in repeater mode, so this is a very annoying issue. I would appreciate someone taking a look.

1 Like

Hi,

Please SSH to the router, and execute the commands:

iwpriv rax0 show MibBucket

logread -f &
while true;do echo ----;iwpriv apclix0 show stainfo;echo ----;iwpriv apclix0 show stacountinfo;echo ----;iwpriv apclix0 show stat;echo ----;iwpriv apclix0 show MibBucket;echo ----;iwpriv apclix0 show mibinfo;echo ----;iwpriv apclix0 show tpinfo=0-2;echo ----;iwpriv apclix0 show vow_info;echo =======;sleep 3;done

When the repeater issue occurs, please copy the issue syslog and share with us.

May I know how do you test the Internet speed of XE3000?

I have saved logs from before, how can I send to you?

Should I run those commands now or after the problem occurs?

I use speedtest, but even loading basic sites like Google is noticeably slow when it happens. And voice calls will keep dropping.

Hello,

you can PM me the syslog through clink the personal icon and message there:

And please help us collect more information:
These commands are running on SSH. When the issue reproduces, please export syslog or copy the content in the SSH window and send it to me.

Is the XE3000 repeater connect to primary WiFi 5G or 2.4G?

If the PC is wired to XE3000, is it normal to access the admin panel and the Internet when this issue reproduced?

2 Likes

I sent you the logs. It depends on the hotel I'm in, but usually 2.4g.

The PC is also connected via wifi. The internet and admin panel remain acessible, but the speed is very slow and the connections often break,

1 Like

Hi,

R&D have checked the syslogs you sent, there is some wireless 2.4GHz interference in your environment.

  1. the channel utilization rate is as high as more than 70%.
  2. Check the clients list, all client devices are connected to XE3000 2.4GHz WiFi. If the repeater is connected to primary router 2.4GHz, and the clients are also connected to XE3000 2.4GHz, it will cause the network speed to decline when the wireless interference becomes larger.

There are several tests can help us to verify:

  1. Please connect the primary router 5GHz if can and your client devices also try to connect to XE3000 5GHz to see if there is improvement for the network speed.
  2. If the hotel does not have 5GHz WiFi, but you can connect your client devices to the XE3000 5GHz WiFi to see if the network speed is stable.
  3. When the issue reproduced, if there is a PC/laptop, please connect to XE3000 through wired cable to test whether the network speed of XE3000 is normal.

Hi,

The 2.4ghz network is disabled. All 2 clients are on the 5ghz network.

The most of the hotel only has 2.4ghz though, so the repeater is on 2.4ghz.

Hi, this is still happening at every hotel I go to.

As I said generally the hotel wifi is 2.4ghz and all the clients are 5ghz.

Hello,

When the 2.4GHz repeater issue reproduced, bring the laptop is wired connect to XE3000, how about the speedtest.net result is low?

I assume that environment 2.4GHz channel is too crowded, which causes the repeater bandwidth to obtain unsatisfactory.

You can do a comparison test, like the XE3000 repeater is connected to the phone hotspot (5GHz) or other primary 5GHz WiFi, and do a speedtest, which can indirectly sense the interference of the 2.4GHz WiFi on that environment.

Hello Bruce,

I don't have an ethernet port so I can't test that directly. I did however run iperf 3 directly from the router to test without the client (iperf3 -c iperf.he.net -4 -R -p 5201 -t 20). Also note this happens almost daily across multiple hotels in multiple countries and is immediately fixed by a reboot, so I do not think it is a crowding issue. Here's my test results:

Problem started:


(ran multiple times, tried various things like disconnect and reconnect client)

iperf3 result:

Immediately after REBOOT:


This pretty much happens daily.

Hi,

Apologize for this question.
The 2.4GHz channel resource utilization rate on site has reached nearly 80%, so it will be slower in network speed.

  1. Please capture the logs again, execute the commands before the issue reproduce, and export the log when it reproduces.
    Please let me know the time point when the problem reproduces, and make sure that this time period is recorded in the log:
logread -f &while true;do echo -----------;iwpriv ra0 show stainfo=1;sleep 3;done
  1. If you confirm that it will reproduce frequency, please share with us the router through GoodCloud. We would like to check this router based on this environment.

Please PM me the router MAC and login Web UI password.


These are the wi-fi signals at my house. -47.5 дБм =943.084 мкВ.
I don't think you have less interference...

@bruce Hi, I can't poll the logs until the error appears because it's sporadic. However, the network issues themselves are frequent but not 100% always reproducible.

I've done a very deep dive into this issue. I do not think this is a congestion issue do to 3 reasons:

  • First of all, there are reports of slow repeater and wifi issues across various models (MT3000, X3000, XE3000) which I've linked a few already. What they have in common is the MT7981 SoC.
  • I experience this on a daily basis across multiple hotels and countries
  • When connecting to hotel wifi directly from my devices there is no issue

I think the key issue is related to:

root@GL-XE3000:~# iwpriv apcli0 stat
apcli0    stat:
RDMode = 0
RDCount = 0
htmode = HE40
CurrentTemperature              = 43
Tx success                      = 69550
Tx fail count                   = 25277, PER=26.6%
Current BW Tx count             = 39668
Other BW Tx count               = 55159
Rx success                      = 455075
Rx with CRC                     = 133087, PER=22.6%
Rx drop due to out of resource  = 0
Rssi: -55 -56 -109 -109
CN Info:                = 0
Last TX Rate                    = MCS13, BW20, 0.4us GI, HT_MM, LDPC
Last RX Rate                    = MCS15, BW40, SGI, HT_MM, BCC
proxy arp enable              = 0
WNMNotify enable              = 0
WNM BSS Transition Management enable = 1
GAS come back delay                       = 0
GAS MMPDU size                            = 0
GAS enable                = 1

There are consistently high transmission failures, which translate into packet loss, dropped tcp connections etc. Eventually, after 24hrs or so, this leads to the following log errors:

7981@C13L1,tx_free_v3_notify_handler() 3530: ContTxFailCntTotal = 1, ContTxFailCnt300ms = 1
7981@C13L1,tx_free_v3_notify_handler() 3533: token used by current wcid = 10, free_token_cnt = 2300

Investigating the driver, I found that the old proprietary mt_wifi driver is being used. It seems the driver cannot recover from transmission failures (doesn't resend), and the resulting poor token pool management eventually leads to the console logs.

I noticed the latest openwrt uses the mt76 driver instead, which also has more features and active development. I hoped that switching to this driver would solve my issues. So I went down the rabbithole and flashed the latest openwrt (24.10.2) and set it up in AP+STA mode to mimic the repeater.

So far this has resulted in significantly less packet loss (as measured by speedtest and packetlosstest.com). However, network instability remains (slow connection start, latency spikes, sporadic slow throughput). Here's an example of a recent Google Meet video call

Checking the new driver stats shows this:

root@OpenWrt:~# iw dev phy0-sta0 station dump | grep tx
	tx bytes:	1764370775
	tx packets:	3117223
	tx retries:	195179
	tx failed:	195186
	tx bitrate:	270.0 MBit/s MCS 15 40MHz
	tx duration:	227510451 us

As you can see there is still a high fail rate. The new driver seems to at least re-transmit, so the packet loss is minimal. That's an improvement compared to the old driver. However, I now believe this is either a deep driver issue, or even a hardware issue.

"There is some issue somewhere in newer mt7x chips that are causing slow downs and many retries."
Here is a recent issue with this driver on the same chipset. The user reports a sudden drop in bitrate in station mode. I polled my device and notice a similar drop:

root@OpenWrt:~# iw dev phy0-sta0 station dump | grep tx
	tx bytes:	1262395487
	tx packets:	1942330
	tx retries:	122582
	tx failed:	122585
	tx bitrate:	135.0 MBit/s MCS 7 40MHz
	tx duration:	140233830 us
root@OpenWrt:~# iw dev phy0-sta0 station dump | grep tx
	tx bytes:	1262405923
	tx packets:	1942365
	tx retries:	122597
	tx failed:	122600
	tx bitrate:	81.0 MBit/s MCS 10 40MHz
	tx duration:	140240368 us
root@OpenWrt:~# iw dev phy0-sta0 station dump | grep tx
	tx bytes:	1262418183
	tx packets:	1942399
	tx retries:	122608
	tx failed:	122611
	tx bitrate:	121.5 MBit/s MCS 6 40MHz
	tx duration:	140245237 us
root@OpenWrt:~# iw dev phy0-sta0 station dump | grep tx
	tx bytes:	1262424943
	tx packets:	1942421
	tx retries:	122618
	tx failed:	122621
	tx bitrate:	54.0 MBit/s MCS 9 40MHz
	tx duration:	140249524 us
root@OpenWrt:~# iw dev phy0-sta0 station dump | grep tx
	tx bytes:	1262453081
	tx packets:	1942501
	tx retries:	122624
	tx failed:	122627
	tx bitrate:	1.0 MBit/s
	tx duration:	140255545 us
root@OpenWrt:~# iw dev phy0-sta0 station dump | grep tx
	tx bytes:	1262953928
	tx packets:	1944033
	tx retries:	122983
	tx failed:	122986
	tx bitrate:	270.0 MBit/s MCS 15 40MHz
	tx duration:	140427978 us

There are many similar reports of issues, such as: mt76->mt7915e: download speed is half upload speed on MT7981 chipset · Issue #980 · openwrt/mt76 · GitHub

So either the MT7981 drivers have been bad for years, I got a bad chip (not about to drop another $500 to test a new one), or the entire MT7981 line of SoCs is defective.

I would appreciate support on this as there are few travel routers that meet my needs (repeater + vpn + 5g cellular + battery), but this router has been frustrating to use.

I did set up GoodCloud, so I can flash back to the stock firmware if you want to test it.