Joined: 03 Jan 2010 Posts: 7568 Location: YWG, Canada
Posted: Sun Oct 28, 2018 0:19 Post subject:
<Kong> wrote:
tatsuya46 wrote:
k4, at least till k4.14 anyway, is trash and should be pulled off kernel.org imo. even broadcom users starting to have our qos problem too apparently.
utter trash.
on the other hand, what about qos cake? brainslayer loves to blindly merge everything openwrt into ddwrt (broken and all), except qos... or we'd have cake. and what about k4.18, .19? anything look different in there for ipq thats worth attempting a port to test?
Cake does not work with imq, if we want to use cake we have to completely redesign qos and most likely also drop a few qos functions we currently ship.
The qos load issue on 4.x is as far as I can tell IPQ only, the R9000 has no issue.
what a shame.. is it possible to use combining queuing dis? like htb+pie on download, htb+fq_codel on upload etc? _________________ LATEST FIRMWARE(S)
BrainSlayer wrote:
we just do it since we do not like any restrictions enforced by stupid cocaine snorting managers
Joined: 15 May 2013 Posts: 243 Location: The Nederlands
Posted: Sun Oct 28, 2018 0:57 Post subject:
<Kong> wrote:
Cake does not work with imq, if we want to use cake we have to completely redesign qos and most likely also drop a few qos functions we currently ship.
The qos load issue on 4.x is as far as I can tell IPQ only, the R9000 has no issue.
So hide the UI until the user selects the QOS he wants to use then show what is supported by the selected? Or gray out the unsupported variables when Cake is selected?
What kind of complete redesign are we talking about? Might it be worth making a single build where there are only 2 options Cake and off? This to test is Cake would be a solution to our problems to begin with.
I know I am asking for a lot right here. _________________ Hi
2x Netgear R7800
B and C is awful bufferbloat. I would like to be at A+ with less than 30 ms and can do on 3.18 only at 500mbs internet. The new 3.18 (37495) is available in main directory _________________ Netgear R7800
Just had serious issues with my throughput, took a while till I found the issue, broken switch, then exchanged it and noticed some qos issues, turns out that the new switch has some EEE features such as low power if it detects low bandwidth. Obviously this feature causes latency issues. _________________ KONG PB's: http://www.desipro.de/ddwrt/
KONG Info: http://tips.desipro.de/
Joined: 03 Jan 2010 Posts: 7568 Location: YWG, Canada
Posted: Tue Oct 30, 2018 11:40 Post subject:
<Kong> wrote:
Just had serious issues with my throughput, took a while till I found the issue, broken switch, then exchanged it and noticed some qos issues, turns out that the new switch has some EEE features such as low power if it detects low bandwidth. Obviously this feature causes latency issues.
have u notified BS about these ipq k4.x issues? this is bullshit, the kernel is broken. certainly something is very wrong here, but i know BS doesnt care about qos, will he revert to k3.18? hes already manually supporting 3.10 on low flash space units so...
u already seen me try to get his attention, hasnt worked ever for qos, and for most other things too, so fed up with this already. maybe he'd listen to u more than us users... _________________ LATEST FIRMWARE(S)
BrainSlayer wrote:
we just do it since we do not like any restrictions enforced by stupid cocaine snorting managers
have u notified BS about these ipq k4.x issues? this is bullshit, the kernel is broken. certainly something is very wrong here, but i know BS doesnt care about qos, will he revert to k3.18? hes already manually supporting 3.10 on low flash space units so...
Is this not an appropriate issue to bring up on the linux-wireless or netdev mailing lists? It sounds like the answer is to get it fixed upstream.
If you know the regression occurred between 4.4 and 3.18 can we not bisect? If you have the bad git commit along with confirming it's broken on 4.19 I'm sure there will be upstream interest in fixing it.
You have a great bufferbloat but probably coz of slow Internet connection.
I still have a twice or even more better bufferbloat on 3.18 compare to 4.9 but on 500mbs _________________ Netgear R7800
14+ days of solid performance on r37495. I do have one issue - the lag when moving from tab to tab seems to be back. It was gone for a few releases but now I occasionally get 10-15 seconds of "hang" when moving to a new tab. _________________ Routing:.......Asus RT-AX88U (Asuswrt-Merlin 384.14) Switching:....Netgear GS608_V3 & GS605_V4, TrendNet TEG-S82G & TEG-S50G
Well damnit!!! The EA8500 just qiut few minutes ago after 16 days
DD-WRT v3.0-r37495M kongat 10/25/18
Linux 4.9.135 #253 SMP PREEMPT Thu Oct 25 02:05:21 CEST 2018 armv7l
Could not get anything to work. No access to it via wired/wireless/webif/SSH/Telnet/
Phone showed still connected to 5GHz Wifi but when I diabaled the phone wifi and tried to reconnect it would not although stupid phone showed it was still broadcasting SSID.
I do occaisionally change port range forwarding to diffent web server if I am working on current web server.
Last time I done that was yesterday morning --- but that aint never caused an issue before.
I just quickly done a power off/on and all is back to normal.
To many people need internet this time of day and many things are linked back to my webserver.
Glad this didn't happen yesterday cause I was gone most that day but connected to its ovpn server all the time without a problem.
Don't have a clue other than something run-a-muck. Guess I need to setup a remote logger....although don't know that it would've caught anything.