Well last night I was only able to try htop. And when I perform the speed test with the CPU monitor I see that it reaches 100%. So, it means that SQM activated. On my Linsys WRT 32X router. Are you running out of processor?
Do as suggested. And I don't change anything. Do you have a symmetrical connection? Apparently with SQM my router is 100% using the processor 
@NeMe_FuUuRyyyyY please post the output of:
cat /proc/interrupts
cat /proc/softirq
one core is maxed out the other idle, maybe the processing could be moved around a bit...
yes, it's quite a bit more sophisticated than my very simple filter. The very simple filter was a simple stopgap for people with 20:1 asymmetric DSL type lines.
I should probably think up some smarter method...
What if we separated acks into their own leaf class with an upper limit of like 10% of the upstream bandwidth, and then put sfq with depth 5 and head-drop below that. Probably smarter than my rate limiting code in nftables.
Why bother? To do it right you need to split the packet into individual flows (or you have to search the full queue for every ACK entering) at which point, just use cake as leaf qdisc...
1 Like
Yeah, I think the only reason to bother would be maybe to have a system that's less CPU intensive than running all the features of cake. But, probably not worth the hassle.
You do have a point, cake has a bad reputation for being resource hungry... I wonder how much of that is caused by the shaper component and how much by the rest, because it is possible to use cake as simple leaf qdisc... it is also possible to overrode cake's priority tier/dscp mappings
Yeah, absolutely if we let HFSC do the shaping and cake be the leaf qdisc with ack suppression and fair queue sort of stuff only it might be a really good idea. Or it might be a bit overpowered. Someone should do the tests comparing cake as leaf qdisc to fq_codel as leaf and see the CPU usage.
1 Like
cat /proc/interrupts
root@OpenWrt:~# cat /proc/interrupts
CPU0 CPU1
25: 0 0 GIC-0 27 Edge gt
26: 92586 42162 GIC-0 29 Edge twd
27: 0 0 MPIC 5 Level armada_370_xp_per_cpu_tick
29: 31626 0 GIC-0 34 Level mv64xxx_i2c
30: 58 0 GIC-0 44 Level ttyS0
40: 0 0 GIC-0 41 Level f1020300.watchdog
44: 0 0 GIC-0 96 Level f1020300.watchdog
45: 86160 0 MPIC 8 Level eth0
46: 0 0 GIC-0 50 Level ehci_hcd:usb1
47: 0 0 GIC-0 51 Level f1090000.crypto
48: 0 0 GIC-0 52 Level f1090000.crypto
49: 0 0 GIC-0 58 Level ahci-mvebu[f10a8000.sata]
50: 95 0 GIC-0 116 Level marvell-nfc
51: 69550 0 GIC-0 57 Level mmc0
52: 0 0 GIC-0 49 Level xhci-hcd:usb2
53: 2 0 GIC-0 54 Level f1060800.xor
54: 2 0 GIC-0 97 Level f1060900.xor
58: 0 0 mv88e6xxx-g1 3 Edge mv88e6xxx-f1072004.mdio-m ii:00-g1-atu-prob
60: 0 0 mv88e6xxx-g1 5 Edge mv88e6xxx-f1072004.mdio-m ii:00-g1-vtu-prob
62: 9 3 mv88e6xxx-g1 7 Edge mv88e6xxx-f1072004.mdio-m ii:00-g2
64: 0 0 mv88e6xxx-g2 0 Edge mv88e6xxx-1:00
65: 0 0 mv88e6xxx-g2 1 Edge mv88e6xxx-1:01
66: 0 0 mv88e6xxx-g2 2 Edge mv88e6xxx-1:02
67: 4 1 mv88e6xxx-g2 3 Edge mv88e6xxx-1:03
68: 5 2 mv88e6xxx-g2 4 Edge mv88e6xxx-1:04
79: 0 0 mv88e6xxx-g2 15 Edge mv88e6xxx-f1072004.mdio-m ii:00-watchdog
80: 0 0 f1018100.gpio 24 Edge gpio-keys
81: 0 0 f1018100.gpio 29 Edge gpio-keys
82: 127783 0 GIC-0 61 Level mwlwifi
83: 44313 0 GIC-0 65 Level mwlwifi
IPI0: 0 1 CPU wakeup interrupts
IPI1: 0 0 Timer broadcast interrupts
IPI2: 2105 2049 Rescheduling interrupts
IPI3: 5524 59080 Function call interrupts
IPI4: 0 0 CPU stop interrupts
IPI5: 0 0 IRQ work interrupts
IPI6: 0 0 completion interrupts
Err: 0
root@OpenWrt:~#
cat /proc/softirq
root@OpenWrt:~# cat /proc/softirq
cat: can't open '/proc/softirq': No such file or directory
root@OpenWrt:~#
root@OpenWrt:~# cat /proc/softirq
cat: can't open '/proc/softirq': No such file or directory
root@OpenWrt:~# cat /proc/softirq
cat: can't open '/proc/softirq': No such file or directory
root@OpenWrt:~# cat /proc/softirq
cat: can't open '/proc/softirq': No such file or directory
root@OpenWrt:~#
Can you help me with the configuration?
Sorry a typo, please try:
cat /proc/softirqs
but /proc/interrupts is already indicating much stuff on CPU0, exactly the CPU that saturated in your htop screenshots...
1 Like
Good perfect! Thanks for the help moeller0. And sorry for the delay, I'm working a lot and I'm late getting home, that's why it takes me a while to respond. Help is appreciated 
Just did a really quick test on an openwrt VM (only one core) and changed:
echo "adding fq_codel qdisc for non-game traffic"
for i in 12 13 14 15; do
tc qdisc add dev "$DEV" parent "1:$i" fq_codel memory_limit $((RATE*200/8)) interval "${INTVL}ms" target "${TARG}ms" quantum $((MTU * 2))
done
to
echo "adding cake qdisc for non-game traffic"
for i in 12 13 14 15; do
tc qdisc add dev "$DEV" parent "1:$i" cake bandwidth ${RATE}kbit ack-filter besteffort
done
Bandwidth was set to:
DOWNRATE=900000
UPRATE=900000
With cake as leaf qdisc for the non gaming classes:
With fq_codel as leaf qdisc for the non gaming classes:
2 Likes
You maybe do not want to use cake's traffic shaper here...
1 Like
unlimited is what you should be using, I believe... if you attach cake to an existing shaper...
2 Likes
Makes sense. Still pretty much the same result with cake:
root@OpenWrt:~# tc -s qdisc
qdisc noqueue 0: dev lo root refcnt 2
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc fq_codel 0: dev eth0 root refcnt 2 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 32Mb ecn drop_batch 64
Sent 33826082638 bytes 22363571 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
maxpacket 0 drop_overlimit 0 new_flow_count 0 ecn_mark 0
new_flows_len 0 old_flows_len 0
qdisc hfsc 1: dev eth1 root refcnt 2 default 13
Sent 151571533 bytes 2253787 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc cake 8184: dev eth1 parent 1:15 bandwidth unlimited besteffort triple-isolate nonat nowash ack-filter split-gso rtt 100ms raw overhead 0
Sent 107879498 bytes 1604442 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 768b of 15140Kb
capacity estimate: 0bit
min/max network layer size: 55 / 87
min/max overhead-adjusted size: 55 / 87
average network hdr offset: 14
Tin 0
thresh 0bit
target 5ms
interval 100ms
pk_delay 0us
av_delay 0us
sp_delay 0us
backlog 0b
pkts 1604442
bytes 107879498
way_inds 0
way_miss 1
way_cols 0
drops 0
marks 0
ack_drop 0
sp_flows 0
bk_flows 1
un_flows 0
max_len 87
quantum 1514
qdisc bfifo 10: dev eth1 parent 1:11 limit 406200b
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc cake 8183: dev eth1 parent 1:14 bandwidth unlimited besteffort triple-isolate nonat nowash ack-filter split-gso rtt 100ms raw overhead 0
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 0b of 15140Kb
capacity estimate: 0bit
min/max network layer size: 65535 / 0
min/max overhead-adjusted size: 65535 / 0
average network hdr offset: 0
Tin 0
thresh 0bit
target 5ms
interval 100ms
pk_delay 0us
av_delay 0us
sp_delay 0us
backlog 0b
pkts 0
bytes 0
way_inds 0
way_miss 0
way_cols 0
drops 0
marks 0
ack_drop 0
sp_flows 0
bk_flows 0
un_flows 0
max_len 0
quantum 1514
qdisc cake 8181: dev eth1 parent 1:12 bandwidth unlimited besteffort triple-isolate nonat nowash ack-filter split-gso rtt 100ms raw overhead 0
Sent 727805 bytes 10829 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 4352b of 15140Kb
capacity estimate: 0bit
min/max network layer size: 55 / 358
min/max overhead-adjusted size: 55 / 358
average network hdr offset: 14
Tin 0
thresh 0bit
target 5ms
interval 100ms
pk_delay 3us
av_delay 0us
sp_delay 0us
backlog 0b
pkts 10829
bytes 727805
way_inds 0
way_miss 5
way_cols 0
drops 0
marks 0
ack_drop 0
sp_flows 1
bk_flows 1
un_flows 0
max_len 358
quantum 1514
qdisc cake 8182: dev eth1 parent 1:13 bandwidth unlimited besteffort triple-isolate nonat nowash ack-filter split-gso rtt 100ms raw overhead 0
Sent 42964132 bytes 638515 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 2534b of 15140Kb
capacity estimate: 0bit
min/max network layer size: 43 / 1309
min/max overhead-adjusted size: 43 / 1309
average network hdr offset: 14
Tin 0
thresh 0bit
target 5ms
interval 100ms
pk_delay 15us
av_delay 1us
sp_delay 0us
backlog 0b
pkts 638515
bytes 42964132
way_inds 0
way_miss 13
way_cols 0
drops 0
marks 0
ack_drop 0
sp_flows 0
bk_flows 1
un_flows 0
max_len 1309
quantum 1514
qdisc ingress ffff: dev eth1 parent ffff:fff1 ----------------
Sent 7512141030 bytes 2675 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc noqueue 0: dev br-lan root refcnt 2
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc hfsc 1: dev ifb-eth1 root refcnt 2 default 13
Sent 7844637471 bytes 5182092 pkt (dropped 34, overlimits 1514854 requeues 0)
backlog 0b 0p requeues 0
qdisc cake 8188: dev ifb-eth1 parent 1:15 bandwidth unlimited besteffort triple-isolate nonat nowash ack-filter split-gso rtt 100ms raw overhead 0
Sent 5620420469 bytes 3712298 pkt (dropped 23, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 1028496b of 15140Kb
capacity estimate: 0bit
min/max network layer size: 1514 / 1515
min/max overhead-adjusted size: 1514 / 1515
average network hdr offset: 14
Tin 0
thresh 0bit
target 5ms
interval 100ms
pk_delay 5.7ms
av_delay 5.54ms
sp_delay 5.24ms
backlog 0b
pkts 3712321
bytes 5620455291
way_inds 0
way_miss 1
way_cols 0
drops 23
marks 0
ack_drop 0
sp_flows 0
bk_flows 1
un_flows 0
max_len 65227
quantum 1514
qdisc cake 8186: dev ifb-eth1 parent 1:13 bandwidth unlimited besteffort triple-isolate nonat nowash ack-filter split-gso rtt 100ms raw overhead 0
Sent 2214701546 bytes 1463137 pkt (dropped 11, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 2009752b of 15140Kb
capacity estimate: 0bit
min/max network layer size: 43 / 1515
min/max overhead-adjusted size: 43 / 1515
average network hdr offset: 14
Tin 0
thresh 0bit
target 5ms
interval 100ms
pk_delay 1.89ms
av_delay 1.72ms
sp_delay 1us
backlog 0b
pkts 1463148
bytes 2214718200
way_inds 2
way_miss 31
way_cols 0
drops 11
marks 0
ack_drop 0
sp_flows 2
bk_flows 1
un_flows 0
max_len 65227
quantum 1514
qdisc bfifo 10: dev ifb-eth1 parent 1:11 limit 406200b
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc cake 8185: dev ifb-eth1 parent 1:12 bandwidth unlimited besteffort triple-isolate nonat nowash ack-filter split-gso rtt 100ms raw overhead 0
Sent 9515456 bytes 6657 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 113016b of 15140Kb
capacity estimate: 0bit
min/max network layer size: 61 / 1515
min/max overhead-adjusted size: 61 / 1515
average network hdr offset: 14
Tin 0
thresh 0bit
target 5ms
interval 100ms
pk_delay 394us
av_delay 186us
sp_delay 1us
backlog 0b
pkts 6657
bytes 9515456
way_inds 0
way_miss 7
way_cols 0
drops 0
marks 0
ack_drop 0
sp_flows 0
bk_flows 1
un_flows 0
max_len 65227
quantum 1514
qdisc cake 8187: dev ifb-eth1 parent 1:14 bandwidth unlimited besteffort triple-isolate nonat nowash ack-filter split-gso rtt 100ms raw overhead 0
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 0b of 15140Kb
capacity estimate: 0bit
min/max network layer size: 65535 / 0
min/max overhead-adjusted size: 65535 / 0
average network hdr offset: 0
Tin 0
thresh 0bit
target 5ms
interval 100ms
pk_delay 0us
av_delay 0us
sp_delay 0us
backlog 0b
pkts 0
bytes 0
way_inds 0
way_miss 0
way_cols 0
drops 0
marks 0
ack_drop 0
sp_flows 0
bk_flows 0
un_flows 0
max_len 0
quantum 1514
root@OpenWrt:~#
1 Like
cat /proc/softirqs
root@OpenWrt:~# cat /proc/softirqs
CPU0 CPU1
HI: 0 0
TIMER: 5194728 2756639
NET_TX: 6557712 63
NET_RX: 20746565 918479
BLOCK: 0 0
IRQ_POLL: 0 0
TASKLET: 63947507 47462
SCHED: 7537011 7011871
HRTIMER: 0 0
RCU: 342399 277834
root@OpenWrt:~#
Same story here, too much work on CPU0, try to enable packet steering and irqbalance...
1 Like
Was wondering how many people are using this script and have felt a distinct effect when playing games?
How do people differentiate between the normal internet and using this script?
Is there gameplay evidence to show theres advantage to using these router based scripts?
Having tired running games with no sqm.
And with sqm enabled i fail to see and difference
Also having used this script also makes no meaning full difference.
The only thing i have felt is when using netem to add delay to ping.
Are these design for people with poor speed, bad connections.
Zia