Hey all,

I play Nexuiz since the 1.2.1 I think. I try all is possible to stop this problem but nothing run.
rate,cl_mouvement, gl_vbo

Problem : Very unstable ping, between 60 and 100 on DCC's servers whereas I had 30 before the server upgrading to 2.2.3 . Seems to be the same problem on all servers tested : PB, RBI, nwn ... but I can't judge them well because I never had good connexion on them.
Since v2.3 release the only server where I can play was the Batserver constant 30 of ping but it was still on 2.2.3. Now it is on 2.3, ping is between 40 and 80.

It is just unplayable for a player who play much and demand good performance.

Other think : I play Nexuiz for DM or 1on1
DM : hard to find a playable and not emptie server / Only minsta ensta ctf or keyhunt now
1on1 : that's ridiculous to play with that 60-100ms unstable ping

That's a pity because I always enjoy playing Nexuiz and the community is very nice.

Cuizi a disappointed player

same here !

with 2.2.3
greens ctf server\dcc delight\pn.de: ping mostly around 1 sometimes 30

tchrs servers: always 30

us server: with good connection 80-120

now with 2.3:
greens ctf server\pn.de: 60-120

dcc delight: 30-90

tchrs servers:120

us servers: 150- 480

that really sucks to play

Friends. Don't look at the ping so much. The connection is still the same, just the value itsself has changed. Don't think your connection has gone worse.

However, I have also my problems with the new version and it's net performance. One day I think "whoa great, runs like hell" and "ffs I'm gonna quit playing Nexuiz right now!".

Fact is that my motivation for playing ladder matches has gone completely byebye. The game performance is too random to seriously play in a sense of competition.

BUT: it really would be unfair to turn your backs to Nexuiz and its developers team just for some problems, instead of helping them for a solution, my 2 cents.

cheers

I don't think i have these problems...but my mouse and keyboard seem to lock up

Urmel.dll wrote: The connection is still the same, just the value itsself has changed.

Not for me man, I really feel the difference during play. For exemple my nex shoots about 0
.5 second after I clicked. Other exemple with shootgun, if the fight is close and if my oponant move, I can't hit him much. All the command answers more than 0.5 seconds after pressed it.
I feel that only on US servers before 2.3, now its everywhere.

Urmel.dll wrote:
However, I have also my problems with the new version and it's net performance. One day I think "whoa great, runs like hell" and "ffs I'm gonna quit playing Nexuiz right now!".

never run like hell on any server, at any hour (tested)

Urmel.dll wrote:
BUT: it really would be unfair to turn your backs to Nexuiz and its developers team just for some problems, instead of helping them for a solution, my 2 cents.

Urm-L =) ! Its not some problems, It's just unplayable, I have not fun anymore playing Nexuiz with that ping...

Maybe something in the network protocol changed that doesn't play well with your router or your ISP, like, it is now filtered or deprioritized because they think it's P2P traffic?

For me, it got better in 2.3, but I am not behind ISPs that filter against P2P traffic. The protocol itself didn't change, but the way the server does rate limiting did, and the times at which the client sends data to the server changed too (you can use cl_netinputpacketsperserverpacket to change the packet rate). It may actually set off some P2P detection routines and then cause a slowdown because of that, some ISPs supposedly do that.

Also, try cl_netinputpacketlosstolerance 1.

I'm also getting a 60 ping on a server that I used to frequent with 20 -30 ping. It isn't a huge difference, but still just thought it was appropriate to mention since the server is fairly close to my hometown.

1on1 : that's ridiculous to play with that 60-100ms unstable ping

Heh, try playing on a German server from where I am...150 +...then come talk to me about shitty performance...I would love 60-100!

IIRC the ping measurement got changed, so the numbers aren't comparable (in the old days people got negative ping values... don't trust the ping values too much). So albeit the displayed number may look different behind the scenes the same stuff happens.

Feel free to toy around with the cl_net* stuff. Apart from the server bandwidth limiting I think you can restore the old behaviour, although I don't know what values would need to get changed.

SavageX wrote:IIRC the ping measurement got changed, so the numbers aren't comparable (in the old days people got negative ping values... don't trust the ping values too much). So albeit the displayed number may look different behind the scenes the same stuff happens.

Feel free to toy around with the cl_net* stuff. Apart from the server bandwidth limiting I think you can restore the old behaviour, although I don't know what values would need to get changed.

why got the ping measurement changed? AFAIK no one said i got a better ping with 2.3

cortez wrote:why got the ping measurement changed? AFAIK no one said i got a better ping with 2.3

Because the old way to measure things wasn't reliable and often showing wrong results.

Also, try cl_netinputpacketlosstolerance 1.

ok, with the cl_netinputpacketlosstolerance 1 seems to be much playable.
always the same high values of ping, but don't feel them.

not good as 2.2.3 but this command run well.
The fact that my ISP filter against P2P traffic is possible because of the TV by adsl which provides.I talked with another french player who have the same ISP, "neuf telecom" and he havn't the 2.3 new lags as me.
my router is a C-box (Sagem Fast 3302) never had problem with it.

Thx for helping me, I think I'll wait a little before quiting =)

I agree with some of these points... theres also some strange server joining lag on the os x ver apparently ( max time to join server ive had is like 7 minutes ) but ingame its still definately playable on some servers

*HxC*Cuizinier wrote:

Also, try cl_netinputpacketlosstolerance 1.

ok, with the cl_netinputpacketlosstolerance 1 seems to be much playable.
always the same high values of ping, but don't feel them.

Good to know it's better for you now.

As for the higher displayed ping value: The actual ping didn't change (after all it's still the same connection), it's just that a more accurate way to measure it is used.

it is very odd that cl_netinputpacketlosstolerance > 0 would affect (improve) movement or decrease felt lag, because it just sends packets twice which, for less good internet connections, will increase (!) ping because the connection is loaded more than before, and I cannot think how it could make movement much smoother (even for my "really" packetloss loaded line the movement-smoothness was not really much affected by the pl-tolerance variable, but, of course switching weapons was working then). Anyhow, a screenshot of your netgraph would really be helpful to tell what is wrong with your connection!

GreEn`mArine wrote:it is very odd that cl_netinputpacketlosstolerance > 0 would affect (improve) movement

Sending it more than once makes it more robust against packetloss. Without that redundance a lost packet of that type would lead to your input getting lost, meaning you'd stop moving for a short fraction of a second on the server, while the client assumes you were moving as usual. This may lead to jerky movement.

Of course one can debate if sending the stuff more than twice makes sense ( > 1) - an already loaded line would be stuffed even further.

Yerstaday, much playable, DCC's 1on1 at 20:05 on game


Today, not much playable, DCC's Minsta at 13:47, before join the game


on game

this time, I really fell the 80 of ping

and you checked if cl_netinputpacketlosstolerance is still 1?

Urmel.dll wrote:and you checked if cl_netinputpacketlosstolerance is still 1?

this one needs to be set on the client-side, doesn't it ?

*HxC*Cuizinier wrote:

Also, try cl_netinputpacketlosstolerance 1.

ok, with the cl_netinputpacketlosstolerance 1 seems to be much playable.
always the same high values of ping, but don't feel them.

Okay, that proves that you do suffer from packet loss.

If you aren't using a WLAN and aren't sharing your internet connection with anyone else, you should call your ISP - it seems to be their fault.

As for the measurements, they now are more correct than before (and less varying).

Cuizi, you should definetely not use the DCC public servers for testing your connection, but just servers that don't modify the sv_clmovement_minping valuie, which are basically all existing v2.3 servers, well, except the DCC public servers. This is because the nice admins of DCC decided to disable client-side movement on their public servers. The DCC 1on1 server however is a good thing for testing because you use your cl_movement there.

If playing on DCC 1on1 is fine but playing on the DCC public servers isn't, it could either mean that it is because of the deactivation of cl_movement on the public server (honetly, if you do have packetloss, playing on a DCC public server will NOT feel good, I tested that myself) or also probably because there are MORE players and your line is more loaded because of that.

And seeing from that screenshot: On the netgraph one can unfortunately only see that you have packetloss on the incoming side. If you have these kind of spikes on server like the 1on1 server it will not be choppy there, because the clientside movement will predict it fine in most of the cases. However, if packetloss also affects your packets (outgoing) TO the server (and this cannot be seen on the netgraph), then your movement will be choppy there, too.

GreEn`mArine wrote:If playing on DCC 1on1 is fine but playing on the DCC public servers isn't, it could either mean that it is because of the deactivation of cl_movement on the public server (honetly, if you do have packetloss, playing on a DCC public server will NOT feel good, I tested that myself) or also probably because there are MORE players and your line is more loaded because of that.

aahh that explains why I've always had to complain about Sxens ser ever's suckiness, they've alvays been carppy for me, and i always have 2-3 or mor PL

GreEn`mArine wrote:you should definetely not use the DCC public servers for testing your connection, but just servers that don't modify the sv_clmovement_minping valuie, which are basically all existing v2.3 servers, well, except the DCC public servers.

Since we moved the servers to the new box (last week or something) they all run with the standard setting for sv_clmovement_minping 0 and sv_maxrate 20000.

Thanks, that's nice to hear. I just assumed that the servers were still set that old way because I didn't test it, and what I said to Cuizi to not use DCC's publics for testing was not a personal issue but a technical.

@green : I test my connexion on DCC's for 2 reason :
- DCC's plain is the only server where you can find the rarely very good DM european players and its one of my two DM playable server.
- DCC's 1on1 is the only 1on1 playable server for me.
even with the improvement of v2.3.
If I can't play on DCC's, I can't play Nexuiz at all because 1on1's is the most funny/interesting for me. (you should know that =) )

@nil : maybe that's client side problems but it's true that DCC's 1on1 is more playable than DCC's plain DM for me