Good idea. Feel free! Incidentally I've a new attempt to get the iptables flavour upstream - it's technically more correct now. https://lore.kernel.org/netfilter-devel/20191203160652.44396-1-ldir@darbyshire-bryant.me.uk/T/#t
This thread been going on a fair bit and wondering if one of you skilled guys can boil it down or refresh the info for 19.07. Like the premise of the multi-quesues and the likes but some of the 1st info implies a loty of unix/iptables knowledge with IPSET and all. I'd love to see a brodies on how to set up...how to configure for you bw up and down and if you wish to prioritise a game/website/subnet its given right there.
Big ask I know but if someone is game, all the kudos!
1 Like
Few Updates: not really seeing huge improvements as expected, not sure if am doing anything wrong. not sure if this method works better for DSL compared to the Cable connection. furthermore even though the DSlReports gives A+ results on Buferbloat, the games don't feel like that at all, take Call of Duty black Ops 2 for example, the connection feels like i am 1 second behind when am in a gun fight with others, when i become a Host, things gets even worse, bullet wont even register while shooting at the player. if anyone have any recommendation with some latest scripts that are more focused on Cable connection, please do share.
This might well be more about your computer's processing power? Not everything is network related. It could also be the game's code. For example when you are host then you would get unfair advantage having maybe 0.01 ms latency compared to others with say 100ms, so they may have some junky code to try to even the playing field, but it doesn't work well?
1 Like
Have you tried a game that can display an in game ping like League of Legends, Overwatch, or Company of Heroes?
I've seen good consistent pings with certain routers, but others will sometimes have brief spikes. Apparently this is due to the propriety wifi driver buffering. It's worse, in my unscientific anecdotal experience, with other devices on my network.
oh this is on Xbox 360 or Xbox one, in both cases the issues is still the same, i have already tried several solution in the past and got helped from various members here and this issue never fully got resolved.
being a host actually makes the other player play better than me and against me, i do much better when am not the host in some cases, its weird but true.
It's not weird if it's a product of either
- Hosting requires a lot more resources so your xbox becomes more bogged down with calculations
- The game code itself tries to penalize you because otherwise you would have an unfair advantage, but the penalizing code over-compensates.
neither of those would be related to network behavior.
Question, the new script defines any $IPT as iptables (IPv4), correct? Am I amble to do IPT="iptables,ip6tables" or something similar so as ipv6 traffic is also included here or is there something else I should do?
Thanks in advance.
edit adding @hisham2630
The current script is for dsl.
I do not have A+ bufferbloat but A only( just because i got jitter through wifi and powerline). I follow @dlakelan on this thread QoS and nftables ... some findings to share. It's also about priority of the packets when congestion is not the issues.
Make sure your bandwidth do not totally saturated.
There is a ton of input lag of your hardware and monitor/tv. I see you play on console and it may connect to a TV which is 60hz? is it a gaming TV 120hz? if not then you may suffer at least 4 frames behind those got 240hz monitor or even worse 1 second behind which is i believe your main issue here. You are still benefit higher hz tv/monitor although the game only 60fps because image don't have to wait and update immediately. When i connect my pc to a normal tv, the latency is extremly high and you may die 4 times before you could react
being host is a good thing on pc. Some still don't like it. You can't be host if you get moderated NAT. you get moderated Nat if you don't port forward the game. Strict Nat is not recommend.
your opponent still shoot back like they don't suffer from aim punch right? It may be because you are lagging behind(monitor/tv latency) and die on the server but on your end, it just a few ms late. Your bullet would not count if you were dead on the server in most game. You may suffer both qos(the script is for dsl) and display latency
I am fully aware of those cases, am not new to this whole thing, I have some knowledge in networking but am not an expert, far as my gaming setup goes, I believe in hard wiring everything if possible and I use monitor for my gaming which have 1ms response time. After playing around with several QOS, SQM, scripts etc I can certainly tell you that so far I haven't found a satisfactory results, that's why I am exploring more solutions to test them out.
Which model? I used to doing fine with my old 120hz 2ms response time and with default sqm piece of cake. 1 second late is a very serious case on pc but i think that common on console because limit of 60fps or less. So if you are sure you are not suffering from network saturation then you should be fine even with piece of cake. I did test a lot with every sqm and qos. Try to cut off internet of everyone and use piece of cake which is most responsive sqm i have ever tested. That way you get a reference of how fast your bullet can register and how responsive you can get. There is no qos or sqm could perfect isolate you from the other user in the same house i'm afraid. I used to be place 1st out of 40000+ players in a private battlefield 4 Chicago server. At that time i just use "sqf" as only qos play through wifi. My accuracy every game was 25 to 30+ percent which is decent and prove there is no packet lost.
Is it only me or there's something wrong!!!, why i don't see latest changes that i made here?
i'm only seeing the old script which is based on ports?!
i can found the changes via search this topic but the main thread is showing old info!?
In the first post I also only see ports based approach, but how about starting a new thread with the newer code and put references here at the end to the last post, and maybe ask @tmomas to close this thread (your choice obviously), since a 700+ post thread might not be the ideal first contact with your scripts...?
4 Likes
Seems great idea my friend!
what about topic name?
Can you make the thread not only dsl but also cable user? There are a few cable users in this giant thread including me
Please! 700+ is a lot to digest.
Not being the OP, I want to add, that the only relevant difference between cable and dsl regarding @hisham2630's scripts are the interface to instantiate sqm on and the specification of the overhead... For cable the per-packet-overhead is typically 18 bytes (22 bytes if a VLAN is in use). And the interface simply is your wan interface... make sure to replace all occurrences of pppoe-wan with what ever your real wan interface is...
1 Like
You said what is in my mind, thank you!
1 Like