Technical info on sockets - suma

[Lt_Dan_Sweden 0]

Hi

To be able to run optimized dedic server and help the community to draw conclusions on Netcode performance, it would be nice to have som technical info on the SOCKETS code. Please provide that.

For example:

Does all of of these still affect performence in SOCKETS?

MaxSizeGuaranteed="512";

MaxBandwidth="2000000";

MaxSizeNonguaranteed="64";

MaxMsgSend="256";

MinBandwidth="900000";

DIRECTPLAY used a couple of ports above 2234. I for example opened 2234 to 2238 to get smooth running. In SOCKETS I am having trouble with ghost players, see previos post

http://www.flashpoint1985.com/cgi-bin....t=15805

This could be that 2302-2306 is not enough?

I have asked before, and will do it again, BIS, please provide a standard method or tool for evaluating the DEDIC server performance. This would help us all in drawing conclusions what is wrong.

My suggestion:

Map: 2-8_T_CAPTURETHEFLAG1.EDEN.PBO

Number of players: 8

Player Max Ping: 250 ms

Ports to open: 2302-XXXX

The SERVER.CFG should look like this LINK_by_BIS

BIS recommended tweaking parameters

MaxSizeGuaranteed="512";

MaxBandwidth="95% of your MAX";

MaxSizeNonguaranteed="64";

MaxMsgSend="256"; ?

Maybe also a tool for catching the #MONITOR 5 messeges into a logfile?

Oh yes, how does the two extra numbers in MONITOR work, packets waiting and packets garanteed?

Regards

Lt_Dan_Dan

[suchey 0]

excellent question...I would like to know this information too.

[russin 0]

you should ask malbouef next time ya see him Suchey..i think he knows a bit about all that jazz come to think of it.

p.s is that Suchey from MTCO server?? if not sorry

oh ya and the guys tell me that you guys think i cheat noway man i don't cheat and never will you can draw your own conclusions i liked playin your server but i feel strange when yas all think im cheating when im just have fun like you guys.

you guys are like my age braket and also good players posees a good challege and fun at the same time but oh well thats the way it go's

[Placebo 29]

"oh ya and the guys tell me that you guys think i cheat noway man i don't cheat" etc.

Russin deal with such things in private, NOT on the forums thanx

[suchey 0]

russin...Ill be sure to ask Mal next time I see him...so far I have just left the config as it was in 1.49 and it seems to be working ok. I do, however, wonder if there are any additional parameters that could be utilized or possibly old ones which arent needed for sockets. Regarding the cheat thing...dont believe what you hear. We have run into a couple of people in the past using the ammo thing, but thats about it...there are always going to be people that have to justify their own lack of ability by accusations. Dont listen to it. Let people talk...its just more fun when you wipe the floor with them that way. Hope to see ya on the server soon!

[Long 1]

Would like to know also, so:

*bump*

[quakergamer 0]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (Lt_Dan_Sweden @ July 16 2002,12:43)</td></tr><tr><td id="QUOTE">Hi

To be able to run optimized dedic server and help the community to draw conclusions on Netcode performance, it would be nice to have som technical info on the SOCKETS code. Please provide that.

For example:

Does all of of these still affect performence in SOCKETS?

MaxSizeGuaranteed="512";

MaxBandwidth="2000000";

MaxSizeNonguaranteed="64";

MaxMsgSend="256";

MinBandwidth="900000";

DIRECTPLAY used a couple of ports above 2234. I for example opened 2234 to 2238 to get smooth running. In SOCKETS I am having trouble with ghost players, see previos post

http://www.flashpoint1985.com/cgi-bin....t=15805

This could be that 2302-2306 is not enough?

I have asked before, and will do it again, BIS, please provide a standard method or tool for evaluating the DEDIC server performance. This would help us all in drawing conclusions what is wrong.

My suggestion:

Map: 2-8_T_CAPTURETHEFLAG1.EDEN.PBO

Number of players: 8

Player Max Ping: 250 ms

Ports to open: 2302-XXXX

The SERVER.CFG should look like this LINK_by_BIS

BIS recommended tweaking parameters

MaxSizeGuaranteed="512";

MaxBandwidth="95% of your MAX";

MaxSizeNonguaranteed="64";

MaxMsgSend="256"; ?

Maybe also a tool for catching the #MONITOR 5 messeges into a logfile?

Oh yes, how does the two extra numbers in MONITOR work, packets waiting and packets garanteed?

Regards

Lt_Dan_Dan<span id='postcolor'>

Good Idea.

Lez wait for BIS to check the thread .

[Lt_Dan_Sweden 0]

BUMP

[grey 0]

why ignore a legitimate question

[Lt_Dan_Sweden 0]

BUMP

[suma 8]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (grey @ Aug. 03 2002,12:32)</td></tr><tr><td id="QUOTE">why ignore a legitimate question<span id='postcolor'>

I cannot see any question in this topic - and that is probably why it is being ignored.

[sgtvor 0]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (Lt_Dan_Sweden @ July 16 2002,12:43)</td></tr><tr><td id="QUOTE">For example:

Does all of of these still affect performence in SOCKETS?

MaxSizeGuaranteed="512";

MaxBandwidth="2000000";

MaxSizeNonguaranteed="64";

MaxMsgSend="256";

MinBandwidth="900000";<span id='postcolor'>

Not a question?

[suma 8]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (sgtvor @ Aug. 12 2002,23:58)</td></tr><tr><td id="QUOTE">Does all of of these still affect performence in SOCKETS?

MaxSizeGuaranteed="512";

MaxBandwidth="2000000";

MaxSizeNonguaranteed="64";

MaxMsgSend="256";

MinBandwidth="900000";

Not a question?<span id='postcolor'>

Yes, all of above is still valid and affects performance in Sockets.

[Espectro (DayZ) 0]

I wonder if we have to ask the questions individually?

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">

DIRECTPLAY used a couple of ports above 2234. I for example opened 2234 to 2238 to get smooth running. In SOCKETS I am having trouble with ghost players, see previos post

http://www.flashpoint1985.com/cgi-bin....t=15805

This could be that 2302-2306 is not enough?

<span id='postcolor'>

[Espectro (DayZ) 0]

And last, but probably not least. Can we have this thingy he is talking about?

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">I have asked before, and will do it again, BIS, please provide a standard method or tool for evaluating the DEDIC server performance. This would help us all in drawing conclusions what is wrong.

My suggestion:

Map: 2-8_T_CAPTURETHEFLAG1.EDEN.PBO

Number of players: 8

Player Max Ping: 250 ms

Ports to open: 2302-XXXX

The SERVER.CFG should look like this LINK_by_BIS

BIS recommended tweaking parameters

MaxSizeGuaranteed="512";

MaxBandwidth="95% of your MAX";

MaxSizeNonguaranteed="64";

MaxMsgSend="256"; ?

Maybe also a tool for catching the #MONITOR 5 messeges into a logfile?

Oh yes, how does the two extra numbers in MONITOR work, packets waiting and packets garanteed?

<span id='postcolor'>

[suma 8]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (Espectro @ Aug. 13 2002,08:43)</td></tr><tr><td id="QUOTE">And last, but probably not least. Can we have this thingy he is talking about?

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">I have asked before, and will do it again, BIS, please provide a standard method or tool for evaluating the DEDIC server performance. This would help us all in drawing conclusions what is wrong.

<span id='postcolor'><span id='postcolor'>

We will not provide any standard method or tool for evaluating. Complexity of creating such tool goes much beyond what we are able to do. If community want do agree on some standard mission that will serve as a kind of benchmark, we have nothing against it  - but server performance is very complicated thing that cannot be tested by single mission and described by signle number.

[suma 8]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (Lt_Dan_Sweden @ July 16 2002,12:43)</td></tr><tr><td id="QUOTE">Oh yes, how does the two extra numbers in MONITOR work, packets waiting and packets garanteed?<span id='postcolor'>

Those two numbers tell how many nonguaranteed and gurannteed messages are waiting on the server because bandwidth is saturated and they have to wait.

[suma 8]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (Espectro @ Aug. 13 2002,08:42)</td></tr><tr><td id="QUOTE">I wonder if we have to ask the questions individually?

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">

DIRECTPLAY used a couple of ports above 2234. I for example opened 2234 to 2238 to get smooth running. In SOCKETS I am having trouble with ghost players, see previos post

http://www.flashpoint1985.com/cgi-bin....t=15805

This could be that 2302-2306 is not enough?

<span id='postcolor'><span id='postcolor'>

There is a lot of other topics about ghosting. There is no reason to think ghosting has any relation to number of open ports.

[grey 0]

i have some questions suma please answer me!

a) is the FPS value given in the monitor thingo capped in any way on dedicated servers in sockets mode, or will it always go as high as it can?

b) when viewing enemy AI soldiers from a long distance away through a scope, their movement seems to "warp" every few metres as they run along in formation. Is this a feature built into the netcode? Eg for objects a certain distance away, they shall be updated less often? Or can certain netcode settings be tweaked to force the server to update the positions of these soldiers more often? We are running on lan and still at long ranges AI soldiers "warp" sometimes entire metres at a time which makes sniping annoying. Can any settings fix/help this?

c) In the admin notes for dedicated server, the setting MaxSizeNonguaranteed is described "Maximum size of non-guaranteed packet in bytes (without headers). Non-guaranteed messages are used for repetitive updates like soldier or vehicle position. Increasing this value may improve bandwidth requirement, but it may increase lag. Default: 256". By improve bandwidth requirement i suppose one means it doesnt require as much. However for a LAN game is it neccessarily good to increase this value? Because if you make this value too big, could the server be "waiting" to get more data to pack into a larger frame, and therefore send less updates of enemy soldier positions? Therefore by this reasoning would it be better to set this to a very small value so the server is forced to send lots of small packets thereby updating soldier position more frequently and eliminating the "warping" effect? Sorry if this last question is a bit newbie/obvious.

Thanks suma

[suma 8]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (grey @ Aug. 13 2002,16:16)</td></tr><tr><td id="QUOTE">a) is the FPS value given in the monitor thingo capped in any way on dedicated servers in sockets mode, or will it always go as high as it can?<span id='postcolor'>

Note: All answers are valid for both DirectPlay and Sockets:

Yes, it is capped - 50 fps.

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">b) when viewing enemy AI soldiers from a long distance away through a scope, their movement seems to "warp" every few metres as they run along in formation. Is this a feature built into the netcode? Eg for objects a certain distance away, they shall be updated less often? Or can certain netcode settings be tweaked to force the server to update the positions of these soldiers more often? <span id='postcolor'>

Yes, it is built in. There is no setting to override this.

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">Because if you make this value too big, could the server be "waiting" to get more data to pack into a larger frame, and therefore send less updates of enemy soldier positions? <span id='postcolor'>

Server is never waiting for updates. It packs only updates generated during one "simulation step" into bigger frames.

[grey 0]

thanks for the replies- so i gather you would recommend larger frames for LAN games ? (and can a setting be "too big" and start causing some kind of side effect? thats what i was really getting at- is there no side effect from large values besides more bandwidth used which doesnt really matter on lan)

And also, as you have said that there is less updates the further away a soldier is, would that not be a good function to implement so that those on good connections/LANs can benefit from their connection? Or would it require massive surgery to the code?

cheers suma

[suma 8]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (grey @ Aug. 15 2002,13:35)</td></tr><tr><td id="QUOTE">thanks for the replies- so i gather you would recommend larger frames for LAN games ? (and can a setting be "too big" and start causing some kind of side effect? thats what i was really getting at- is there no side effect from large values besides more bandwidth used which doesnt really matter on lan)<span id='postcolor'>

I think on LAN you could increase this to 1024. Increasing it further has no sense, as currently maximum allowed packet size in our socket implementation is 1490 minus headers. I am not aware of any sideeffects. Note: while this setting affects server most, as server traffic is usually highest, it is valid on both server and clients. Each computer has its own settings and limits which affect OFP outgoing network traffic. (MaxSizeGuaranteed, MaxBandwidth, MaxSizeNonguaranteed, MaxMsgSend and MinBandwidth settings).

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">And also, as you have said that there is less updates the further away a soldier is, would that not be a good function to implement so that those on good connections/LANs can benefit from their connection? Or would it require massive surgery to the code?

<span id='postcolor'>

I think adjusting this should be pretty easy. We will consider it for future patches.

[Espectro (DayZ) 0]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">patches<span id='postcolor'>

Yay!

There will be multiple!!

[hees 0]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (Hees @ Aug. 25 2002,16:26)</td></tr><tr><td id="QUOTE"><span id='postcolor'>

Is there one definitive list that describes all there is to server .cfg's and internet/LAN multiplay?

[benu 1]

I'd like to see an extended server docu too.

And where can i setup networking options for my client? I just checked my Flashpoint.cfg and found no settings for networking?!?

Regarding max packet size (1490 - headers): How much are the headers? 4 bytes? 8 Bytes? Or can i set max packet size to 1500 and ofp will cap it at the max whatever it is? I guess i have to set it to lower value because of the tcp/ip and pppoe headers anyway....