Hm could you show your SQM settings please?
1 Like
Sorry for the delay
root@OpenWrt:/etc/config# cat sqm
config queue 'eth1'
option qdisc 'cake'
option interface 'br-lan'
option download '0'
option debug_logging '0'
option verbosity '5'
option qdisc_advanced '1'
option qdisc_really_really_advanced '1'
option linklayer 'none'
option eqdisc_opts 'diffserv8 nat datacentre'
option upload '15000'
option ingress_ecn 'NOECN'
option egress_ecn 'NOECN'
option script 'layer_cake.qos'
option squash_dscp '0'
option squash_ingress '1'
option enabled '1'
config queue
option download '0'
option debug_logging '0'
option verbosity '5'
option qdisc 'cake'
option qdisc_advanced '1'
option qdisc_really_really_advanced '1'
option interface 'pppoe-wan'
option eqdisc_opts 'diffserv8 nat ack-filter datacentre'
option linklayer 'none'
option upload '15000'
option ingress_ecn 'NOECN'
option egress_ecn 'NOECN'
option script 'layer_cake.qos'
option squash_dscp '0'
option squash_ingress '1'
option enabled '1'
That is not really ideal...
and datacentre (5/100 µs) is not going to work any better than lan (50/1000 µs) and I doubt that metro (0.5/10 ms) is going to be significantly worse... at 15000 Kbps a single full MTU packet requires something like
(((1500 Byte+38 Byte ) * 8 bits/Byte) / 15000000 bis/sec) * 1000 ms/sec= 0.82 ms
since cake defaults to 1.5 times a full MTU target you will result with a ~0.82*1.5 = 1.23 ~ 1.5ms ms target (and given that you use diffserv8 probably larger) and a minimum interval of ~3ms...
Is that still true if you use regional instead of datacentre?
The firmware i am using is 21.02 with DSA, should i put instead of br-lan switch port lan1 where i connect my console? I tried now putting regional but upload remains about half
Probably not... because you need to shape all of your traffic (if you want to control bufferbloat that is, if all you want is to limit the consoles maximum rate instantiating a shaper in lan1 would be fine, but I think that is not what you use these scripts for...)
For swconfig we always had a "handle/name" for the interface from the CPU to the switch, but for DSA I believe that disappeared and so br-lan might be the best you can get.
For testing purposes, could you disable your current SQM instances and instantiate a new one for testing on the wan interface with both ingress/egress rate set to 15000, please? The idea is first to establish whether the issue can be recreated with a simpler set-up....
And if you use the default internet keyword, what speeds do you measure then?
1 Like
I may be way off here, but... here it goes:
Does setting jumbo frames on on the OperWrt Router itself, or on managed switches have any impact (positive, negative) on what we are discussing here?
As always, thanks in advance for anyone putting out words of wisdom.
Cesar
Jumboframes only ever help if all computers involved in your traffic are configured to use jumbo frames, and typically the internet is restricted to ethernet's default maximum packet payload size of 1500 Bytes.
2 Likes
hi everybody how found the latency server like this
he work only for warzone for the moment
ipset=/d3ovluux6b7f2q.cloudfront.net/*demonware.net/Iw8-ps4-loginqueue.prod.demonware.net,3074,1/latsens
this packets like this
i would like battefield and cold war
1 Like
this morning I tried to create a single instance in sqm with upload and download at 15000 without other options but the upload is always half, maybe sqm does a test and finds that my line is not symmetrical (my connection is fttc 55down 20up)? strange
update i did other tests with separate instances on lan and wan and it seems that the problem is the dslreports test site, exactly connection type option, only by selecting gigabit the values are correct (even if i don't have a gigabit) .thanks @moeller0 for help
Well, is this also true if you use the internet keyword instead of datacentre?
I tried it's true even with the internet
Okay, could you follow these instructions for how to configure a dslreports speedtest and how to link to the detailed results page please?
And then run the speedtest with say No. upload streams: 4 and No. upload streams: 16, then we might be able to see something in the retransmit rates and the aggregate throughput.
Okay, so it is not necessarily cake's aggressiveness in dropping...
result with No. upload streams: 4 and option datacentre
result with No. upload streams: 16 and option datacentre
Looking at these results the really small congestion windows pop out, so basically the servers allow to little data in flight to fill the pipe, and hence the throughput per flow is pretty low, also in the reported Re-xmit Avg values are really high (which fits with the low cwnd).
The test looks quite symmetrical to me though so I fail to see your reported issue, what am I missing? Also could you repeat this with internet instead of datacentre and post the results....
And finally, could you post the output of tc -s qdisc first with datacentre then with internet, please?
hello everyone, i'm back and motivated to test your talent as a network engineer 
indeed i have moved and i have a new lower speed connection, vdsl2 70/20 .. i play mainly fifa and since a few months at COD .. I am on PS5 console .. I have a latency which does not allow me to compete against my opponents, the experience of the game is ruined despite my new home and new connection. I regret to see that I am not playing in real time, because I am not receiving the image of the game in real time .. I would like to test the script, I am a little confused I do not know what the procedure is and the last one script to configure .. could you guide me through this process? thank you in advance for the help and the time you will be available to give
I compare a dslreports result with a default configuration and a new test with your configuration, I see a huge difference between the default config which gives me an excellent result in appearances compare to your configuration which gives me an almost catastrophic result which corresponds much more to the reality of a part of cod .. thank you for paring this method
this is the standard test with asymmetric values
the problem occurs only using the standard test on dslreport.
Do you also want me to send you the other tests? And qdisc output?
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 1518 ta rget 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
Sent 48971190 bytes 170413 pkt (dropped 0, overlimits 0 requeues 4)
backlog 0b 0p requeues 4
maxpacket 3466 drop_overlimit 0 new_flow_count 109 ecn_mark 0
new_flows_len 0 old_flows_len 0
qdisc noqueue 0: dev lan1 root refcnt 2
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc noqueue 0: dev lan2 root refcnt 2
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc noqueue 0: dev lan3 root refcnt 2
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc noqueue 0: dev lan4 root refcnt 2
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
qdisc cake 800c: dev wan root refcnt 2 bandwidth 12Mbit diffserv4 dual-srchost n at nowash no-ack-filter split-gso rtt 100ms noatm overhead 44
Sent 8427580 bytes 74695 pkt (dropped 11, overlimits 5726 requeues 0)
backlog 0b 0p requeues 0
memory used: 53586b of 4Mb
capacity estimate: 12Mbit
min/max network layer size: 28 / 1500
min/max overhead-adjusted size: 72 / 1544
average network hdr offset: 14
Bulk Best Effort Video Voice
thresh 750Kbit 12Mbit 6Mbit 3Mbit
target 24.2ms 5ms 5ms 6.05ms
interval 119ms 100ms 100ms 101ms
pk_delay 2us 2.8ms 55us 253us
av_delay 0us 227us 8us 10us
sp_delay 0us 2us 2us 2us
backlog 0b 0b 0b 0b
pkts 3 17549 809 56345
bytes 162 3251063 48090 5137945
way_inds 0 73 2 0
way_miss 2 1073 408 52
way_cols 0 0 0 0
drops 0 11 0 0
marks 0 0 0 0
ack_drop 0 0 0 0
sp_flows 0 1 4 1
bk_flows 0 1 0 0
un_flows 0 0 0 0
max_len 54 6056 90 3028
quantum 300 366 300 300
qdisc ingress ffff: dev wan parent ffff:fff1 ----------------
Sent 33205814 bytes 87214 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 cake 800d: dev ifb4wan root refcnt 2 bandwidth 57Mbit diffserv4 dual-dstho st nat nowash no-ack-filter split-gso rtt 100ms noatm overhead 44
Sent 34599546 bytes 87205 pkt (dropped 9, overlimits 25795 requeues 0)
backlog 0b 0p requeues 0
memory used: 88944b of 4Mb
capacity estimate: 57Mbit
min/max network layer size: 46 / 1500
min/max overhead-adjusted size: 90 / 1544
average network hdr offset: 14
Bulk Best Effort Video Voice
thresh 3562Kbit 57Mbit 28500Kbit 14250Kbit
target 5.1ms 5ms 5ms 5ms
interval 100ms 100ms 100ms 100ms
pk_delay 3us 291us 22us 110us
av_delay 0us 110us 3us 6us
sp_delay 0us 2us 3us 3us
backlog 0b 0b 0b 0b
pkts 5 86544 193 472
bytes 366 34514989 43914 53903
way_inds 0 310 0 0
way_miss 4 1618 17 50
way_cols 0 0 0 0
drops 0 9 0 0
marks 0 0 0 0
ack_drop 0 0 0 0
sp_flows 1 3 0 1
bk_flows 0 1 1 0
un_flows 0 0 0 0
max_len 98 19682 1514 4452
quantum 300 1514 869 434
root@OpenWrt:~#