All Activity

Might be useful for someone, looks like RFC4638 is supported (tested in 4.3), the only thing you need to set MTU using CLI: interface ISP ip mtu 1508 system configuration save When PPPoE connection is restarted I am able to increase ICMP packet size up to 1472 (which corresponds to MTU 1500 minus 28 bytes header) The only drawback is that UI settings show error for the field, if you need to change any setting on this page, it will not let you save it. The workaround is to reduce it to 1500 in UI, save, and change using CLI again. This could be something for Keenetic team to consider fixing (bump valid value to MTU size field to 1508).

Hello, my qos does not detect my games as gaming. Can I modify qos and add my game ports so that I prioritize them correctly??

22/01/2026 Keenetic RMM System requirements change Keenetic RMM now requires the controller to run OS version 4.1 or higher. Raising the minimum version helps ensure security, stability, and high performance.

Had same problem. Some context: Device registered in Russia. Installed Keenetic OS language was English Next steps helped me to solve it: In General System Settings -> Component Options install Russian language Reboot device and accept some new agreement Now acquire works I guess accepting of this agreement is required for acquire operation

In order for IntelliQoS to properly prioritize traffic it needs to know the total amount of bandwidth available, that's why you see different results after manually specifying the download and upload speed. You said before that without any IntelliQoS configuration you typically get bandwidth values around 92 Mbps (download) / 22Mbps (upload). How do you measure those values? I am asking, because you said that you see different bufferbloat results after entering lower values manually (85/20). Could it be that the lower numbers are closer to the actual bandwidth of your connection?

I have huge prob;ems using my ssd as storage device for qbittorrent(had the sma eproblem with the default transmission client). The router just cand t handle devices with that kind os specs (the manufacturer confirmed that the port provides less power than need, 900mA or so). I tried a y usb cable, with aditional power source, it is much bettter this way, but still have issues from time to time, my router disconnects randomly my ssd storage). Have anyone a (real) solution for this problem? We are talking about a Titan 1811 2nd gen router.

I'm just experiencing the new router I bought. I was expecting IntelliQoS to work without limiting my overall network speed, but it seems not to. Bufferbloat test results are same if I don't touch SQM parameters. So what are the other options useful for? The priority options. I was trying to achieve better bufferbloat tests than my previous router, but without limiting network speeds of devices.

Hi, @reshuffle812 What problem are you trying to solve with IntelliQoS? I see that you check the configuration by running a bufferbloat test, but if you could elaborate on what you're trying to achieve with getting certain values there, maybe we can provide a better advice about how to configure the Keenetic.

I updated my router firmware to 5.0.4 The problem is solved Thank you.

First time Keenetic user here. I enabled Traffic Classification Engine, and I have a theoretical 100Mbps/20Mbps connection as download and upload, although I'd usually get 92Mbps/22Mbps, and after having a Keenetic that's around 87Mbps/22Mbps now. I'm not sure if that's caused by Keenetic router, or the cheap Ethernet switch which supports up to 100Mbps. Anyways, I'm not exactly sure how should I use IntelliQoS. When I only enable Application Classification and Application Prioritization (which auto enable the other) options it does nothing helpful, Bufferbloat tests show the same. When I enable Smart Queue Management in the same page and set it to 85Mbps/20Mbps, bufferbloat tests gets better. Basically bufferbloat during download is up to 10ms, and for upload it's 0ms or 1ms at max. Even better than download, although I have more download to spare, but somehow bufferbloat during upload became better. So is this the only way? I'm not sure if it makes any sense like this, my previous router was TP-Link, given by my ISP, and as you know ISP given (or rented) routers have an interface controlled by ISP, most features limited, but even there limiting QoS per device would give me the similar results. I was expecting Keenetic's anti-bufferbloat solutions would work better because I have more control on that router, am I wrong thinking that? Or did I do something wrong in the setup? Forgot to include main concern here, main concern is that limiting network speeds manually like that, I have seen some OpenWrt users talking before and I didn't heard they have to do this network limit.

NDM-4235

Release 5.0.4 (preview): CIFS: fixed compatibility with MacOS TimeMachine [NDM-4112] gnutls: updated to 3.8.11 [NDM-4205]: CVE-2025-32988 CVE-2025-32989 CVE-2025-32990 CVE-2025-6395 CVE-2025-9820 HTTP: fixed loss of access to Web UI due to network discovery timeouts [NDM-4223] IP: fixed adding of static routes after the default route [NDM-4209] NTCE: optimized memory consumption [NDM-4219] SSDP: fixed system crash during network discovery [NDM-4224] Web: added Name sorting to DNS-Based Routing [NWI-4595] Web: fixed configuration saving without VLAN ID (reported by @Mikhail_YAR) [NWI-4578]

The modem currently operates in USB serial mode. This results in low speed and limited functionality. I'm trying to get USB composition with QMI so I can add support for the correct mode. Please send more sys:/kernel/debug/usb/devices or self-test file.

I wrote that in the initial message. By default the modem is in mbim mode: Interface bitmask: 00001009 (diag,modem,mbim) In that mode the connection is tried to initialize but always fails. Once mode is changed to rmnet0: Interface bitmask: 00000109 (diag,modem,rmnet0) The modem gains the bare minimum functionality with no working AT sending and no IPv6.

The necessary corrections have been made in the versions 4.03.C.7.0 and 5.00.C.4.0

A year has passed. The problem remains unresolved. What should I do?

Hello, please, check the telephone line settings on your Skipper, they should be as on the "skywifi-telephone-line-settings-20260112.png" screenshot below. Enable the 100rel option by sending the command “nvox sip-common 100rel” and “copy running-config startup-config” (see other screenshots below). Read the article for learning about Keenetic command-line interface. After enabling the 100rel option, try incoming and outgoing calls again.

Does this modem support QMI (rmnet) USB composition? Please switch the modem to QMI and get info about composition: more sys:/kernel/debug/usb/devices

Приветствую. У меня это получилось победить только благодаря изменению крипта S92syncthing на запуск в режиме демона. Возможно Вам тоже поможет. Первым делом создаем пользователя через adduser под запуск Syncthing - у меня это sync_user. Потом меняем права у папки /opt/etc/syncthing (если у Вас другая, то указывайте свою) и файлов в ней с помощью chmod: chmod -R /opt/etc/syncthing После этого вносим изменения в скрипт /opt/etc/init.d/S92syncthing: PROCS=syncthing ARGS="-- --home=/opt/etc/syncthing" PREARGS="start-stop-daemon -S -c sync_user:sync_user -x" Если кратко, то в скрипте /opt/etc/init.d/rc.func - это все превращается в команду: start-stop-daemon -S -c sync_user:sync_user -x syncthing -- --home=/opt/etc/syncthing Что соответственно запускает скрипт от имени простого пользователя. Проверено на актуальной версии syncthing на момент публикации (1.29.6-1).

In KeeneticOS 5.0+ a very useful feature was added — domain name based routing (DNS-based routes). However, there is a fundamental issue: the first packet is always sent via the default route. Current behavior: • a client starts a connection • the first TCP SYN is sent via the default interface • only after the DNS response Keenetic adds the routing entry • the second attempt goes through the correct interface As a result: • the first request is often slow or times out • many websites and apps “open on the second try” • network behavior becomes unpredictable Expected behavior: If a domain is listed in domain name routing rules, the very first packet should be routed via the correct interface, not the default one. Suggestion: Implement a mechanism where Keenetic waits for the DNS response and applies the route before sending the first packet, so the connection is correct from the beginning.

(config)> ip dhcp pool _WEBADMIN option 66 ascii 192.168.2.1 (config)> system configuration save with second row, its ok now, thanks a lot

It works, but after restart the config values gone,

Thanks a lot, I ll try it tomorrow at the office,

If you press Tab after entering part of the command it will show you the available autocomplete options. For example, if you press Tab after entering ip dhcp pool<space>, you will see the names of existing DHCP pool objects. After that you can check which options are available after you enter ip dhcp pool __POOL_NAME__ option<space> or ip dhcp pool __POOL_NAME__ option 66<space>. Maybe that will help you.

same... (config)> ip dhcp pool _WEBADMIN option 66 string 192.168.2.1 Command::Base error[7405602]: argument parse error.