[OpenWrt-Devel] Tried LEDE with SQM today

[Maciej Soltysiak]

Right, so it seems it's technically in-there, maybe just not performing...
Here's an lsmod and 2 tc -s qdisc show dev eth1|ifb4eth1 showing no badness
me thinks.

root at openwrt:~# lsmod|grep ^sch_cake
sch_cake               16071  0

root at openwrt:~# tc -s qdisc show dev eth1
qdisc cake 802a: root refcnt 2 bandwidth 3584Kbit besteffort flows rtt
100.0ms raw
 Sent 12207688 bytes 68580 pkt (dropped 41, overlimits 128117 requeues 0)
 backlog 0b 0p requeues 0
 memory used: 790400b of 4Mb
 capacity estimate: 3584Kbit
                 Tin 0
  thresh      3584Kbit
  target         5.1ms
  interval     100.1ms
  pk_delay      30.7ms
  av_delay      10.3ms
  sp_delay        12us
  pkts           68621
  bytes       12211162
  way_inds           0
  way_miss         120
  way_cols           0
  drops             41
  marks           2507
  sp_flows           6
  bk_flows           1
  un_flows           0
  max_len         1514
qdisc ingress ffff: parent ffff:fff1 ----------------
 Sent 78688860 bytes 67891 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0

qdisc cake 802b: root refcnt 2 bandwidth 85Mbit besteffort flows rtt
100.0ms raw
 Sent 79761689 bytes 68235 pkt (dropped 0, overlimits 62444 requeues 0)
 backlog 0b 0p requeues 0
 memory used: 388864b of 4250000b
 capacity estimate: 85Mbit
                 Tin 0
  thresh        85Mbit
  target         5.0ms
  interval     100.0ms
  pk_delay       258us
  av_delay        22us
  sp_delay         6us
  pkts           68235
  bytes       79761689
  way_inds           0
  way_miss         129
  way_cols           0
  drops              0
  marks             17
  sp_flows           0
  bk_flows           1
  un_flows           0
  max_len         1514




On Sun, Nov 13, 2016 at 11:25 PM, Dave Taht <dave.taht at gmail.com> wrote:

> kmod-sched-cake is needed
>
> tc -s qdisc show your_device would probably show absence or badness.
>
> I do not think that hardware is capable of 90mbits of inbound shaping
> either.
>
> On Sun, Nov 13, 2016 at 2:23 PM, Maciej Soltysiak <maciej at soltysiak.com>
> wrote:
> > Hi,
> >
> > I gave LEDE a whirl on WNDR3800 on a 100 Mb DSL line.
> > Without SQM, speedtest.net gave around 90 Mbit/s of download and 3,6
> upload.
> > I installed sqm-scripts and the luci app and enabled SQM on eth1, set
> > ingress and egress shaping to 90% of speedtest.net results, enabled
> fq_codel
> > and simple.qos
> >
> > Now when on to dslreports bufferbloat tests and got about 42 Mbit/s.
> Great
> > bufferbloat result but ingress seems to perform at just above 50% of
> ingress
> > threshold (42 / 90% * 90).
> >
> > I don't think I've seen that drastic effect when I was on cero and then
> on
> > openwrt.
> > Anything obvious I'm doing wrong?
> >
> > Also, funny thing, there's a sqm-scripts-extra with cake and also the
> > packages seem to contain the right kernel modules, but there's not cake
> > queueing disciple in the luci UI. I found in
> /var/run/sqm/available_qdiscs
> > empty files which correspond to the dropdown in luci: codel, fq_codel,
> pie,
> > sfq. Could it be that it's just missing an empty file for cake or is that
> > something more deep?
> >
> > Best regards,
> > Maciej
> >
> >
> > _______________________________________________
> > openwrt-devel mailing list
> > openwrt-devel at lists.openwrt.org
> > https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
> >
>
>
>
> --
> Dave Täht
> Let's go make home routers and wifi faster! With better software!
> http://blog.cerowrt.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.infradead.org/pipermail/openwrt-devel/attachments/20161113/a4a2560f/attachment.htm>
-------------- next part --------------
_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel