[FS Frankie V]

vareul, on 01 February 2014 - 06:09 AM, said:

Ive been having very severe forms of the issues specified in this thread.

Check with the server operator and have them check to make sure that there is no 4.1 based asset pk3 files placed in the games directory.

Using 4.1 control and ini files can mess with server side hits in 4.2.

[DragonHeart]

Frankie V, on 01 February 2014 - 06:47 AM, said:

Check with the server operator and have them check to make sure that there is no 4.1 based asset pk3 files placed in the games directory.

Using 4.1 control and ini files can mess with server side hits in 4.2.

My 4.2 server was a complete from-scratch implementation and has no way of reading any assets from the old 4.1 server. There is no chance of foreign .pk3s (other than a limited number of maps) contaminating the server. If there is a problem it is elsewhere.

[vareul]

DragonHeart, on 01 February 2014 - 06:57 AM, said:

My 4.2 server was a complete from-scratch implementation and has no way of reading any assets from the old 4.1 server. There is no chance of foreign .pk3s (other than a limited number of maps) contaminating the server. If there is a problem it is elsewhere.

Thats what i was thinking. I dont really think its FA's servers. I doubt that it is. Theres people from various servers and clans that have mentioned having this same issue.

I dont think i should hit every shot i take, nor do i think im a marksman level player. I just find it a bit odd that i miss so often in these very controlled shots.

[FS Frankie V]

DragonHeart, on 01 February 2014 - 06:57 AM, said:

My 4.2 server was a complete from-scratch implementation and has no way of reading any assets from the old 4.1 server. There is no chance of foreign .pk3s (other than a limited number of maps) contaminating the server. If there is a problem it is elsewhere.

Using the majority of server reports collision detection is working the way it should.

Backtracking in 003 there was a problem discovered, assuming a clean from scratch install, that if a pk3 file was corrupted then that pk3 was ignored and would roll back to the last packaged loaded that contained the required control file that does not match the asset file in question.

To solve this problem pk3 files were reordered and matching control and ini files removed.

Based on the symptom as described the only “known” way this can occur is either a 4.1 asset package was added to the 4.2 game folder or one of the main pk3 files was corrupted and ignored during game load.

In this case that would be the control file for the player animations.

If it is a case of corruption then the fix would be to delete zUrT42_0013.pk3 and re-download it to see if it solves your problem.

If not then further testing would be required and if you can run the asset package directly from the server in local mode you could visually check to see if the player models and hitmesh are still glued together.

Last check all of the pk3 files. Only zUrT42_0013.pk3 should contain models/players/<model_name>/animation.cfg and if you find any other packages containing this control file let us know as soon as possible.

It should also be noted that the update to the hitmesh only corrects for collision detection problems in 4.1 and p5yc0runn3r is working on the second half of the problem.

[vareul]

Whats the second half of the problem?

I dont mean this in any pushy or agitated way. I know what i encountered. I didnt count anything as odd if it wasnt in as controlled a test as i could get in an active CTF server. It might take me a half hour or more to run into someone whos camping a corner and doesnt kill me in a half second once im in line of sight. Only thing recording my gameplay would show is that my crosshair and rate of fire are certainly well controlled in those situations.

Theres a marked difference between missing a moving target because of ping or spread, and missing a stationary target while crouched, and firing in semi auto mode. My shots are rather tightly grouped in full auto mode, for the first 5-6 bullets, when shooting at a wall. Of course range effects the spread of burst fire, even more so for full auto. This issue might just be one born of the way the game handles traffic. Something about that doesnt seem right, though. I highly doubt that my average of 80 ping, would cause my stationary, crouched shots, to miss 5 out of 8 times, since the target was AFK in a corner. It took me a good 12 seconds to kill the person. They didnt even turn around.

Im aware that noones trying to hide anything, thats not what im saying. Im just saying that, logs aside, im having a real issue.

This post has been edited by vareul: 01 February 2014 - 06:26 PM

[DragonHeart]

@ Frankie

I verified the md5sum of all .pk3 files and they were all correct, so if it is corrupt, it's corrupt upstream.
Additionally, I looked through all .pk3 files to verify that the player models are only present in zUrT42_0013.pk3 and this is indeed the case, so asset duplication doesn't appear to be the problem, at least from an end-user perspective.

[FS Frankie V]

vareul, on 01 February 2014 - 06:23 PM, said:

Whats the second half of the problem?

Im aware that noones trying to hide anything, thats not what im saying. Im just saying that, logs aside, im having a real issue.

Since you asked so nicely I'll do one better.

Since psyco is going into beta testing I'll put together a design document over the next few days based on some kind of a time line as to what was fixed and the reasons as to those fixes so thous going into the testing are up to speed that this is not just a change but something that has to be done.

I'll not be covering what will be tested as I know psyco has his own plans laid out and will explain rhyme and reasons as to these changes when he is ready.

For the record we don't hide anything, we only hold on to a few things until they become relevant.

DragonHeart, on 01 February 2014 - 11:51 PM, said:

@ Frankie

I verified the md5sum of all .pk3 files and they were all correct, so if it is corrupt, it's corrupt upstream.
Additionally, I looked through all .pk3 files to verify that the player models are only present in zUrT42_0013.pk3 and this is indeed the case, so asset duplication doesn't appear to be the problem, at least from an end-user perspective.

The way the connection works the client manages the visible side of content and the server manages all things relating to collision detection. In all cases the server is 100% right as to the outcome of a given event.

It does not go client > sever > client > hit

It goes client > server > question was there a hit?

The client will always see if there is a mismatch of some kind, yah know moving around a map with your butt on the ground, and is hard coded in on the server side that can not change based on what the client sees.

My recommendation at this time is to scrub your server clean and rebuild from scratch using only what is available from our down loader. If the problem persists contact me by PM, with your server info, as at that point it does become concerning as to the root of the problem and I'll see if I can get a few of the guys, and myself, and we will head over to see what we can see.

The bottom line all things being equal the results of all testing done has always backtracked to something wrong on the server side as we have tested under the worst conditions possible which is not to say a problem cannot appear that was not anticipated.

[thelionroars]

While reinstalling the server might be a very methodical way of troubleshooting, if it is a fresh install and the md5s have been checked, I'm very doubtful it is going to fix anything. You could try running through the server configs to check if anything is there that shouldn't be I suppose.

If I'm not mistaken I was the player in the crosshairs when vereul experienced this issue the other day. I think I ran out of bullets, reloaded and then finished him off while he shot me with all his weapons and didn't get a hit. It did seem very odd and that he should have hit me. I experienced something similar playing a 1v1 on a local server with another low-ping player a few days ago. I aimed at them and shot them before they'd seen me, and there were no hits at all when it seemed that I would have damaged them at the very least. On the same server a day or two earlier, up to ten players and no one mentioned any problems. I think there needs to be a consistent pattern across many servers and players to state outright that there is an issue, but I will say I have seen a couple of incidents that seemed odd lately.

[FS Frankie V]

Well the more information the better but we do have to account for strange as compared against the average.

However the hitmesh design is now on the very same channels that controls the visible player models and is wired in just like the LOD models. Ping is irrelevant at this scale and is unique as to reason why this would occur and is why it is of concern.

What is a red flag is the high number of non-collisions occurring in the crouch position that a simple test we do is to fire at where the torso or head would be if the player is standing up. If you get hits then it's a control file miss match between the client and server.

The next test would be to have player A (cheats and God turned on) stand in a fixed position, moving to a crouch position, walking, running while player B fires at and around the player to pattern where the hitmesh is in relationship to the visible player model.

[vareul]

thelionroars, on 02 February 2014 - 02:19 AM, said:

While reinstalling the server might be a very methodical way of troubleshooting, if it is a fresh install and the md5s have been checked, I'm very doubtful it is going to fix anything. You could try running through the server configs to check if anything is there that shouldn't be I suppose.

If I'm not mistaken I was the player in the crosshairs when vereul experienced this issue the other day. I think I ran out of bullets, reloaded and then finished him off while he shot me with all his weapons and didn't get a hit. It did seem very odd and that he should have hit me. I experienced something similar playing a 1v1 on a local server with another low-ping player a few days ago. I aimed at them and shot them before they'd seen me, and there were no hits at all when it seemed that I would have damaged them at the very least. On the same server a day or two earlier, up to ten players and no one mentioned any problems. I think there needs to be a consistent pattern across many servers and players to state outright that there is an issue, but I will say I have seen a couple of incidents that seemed odd lately.

You were one of them. However the issue i had against you included factors of movement and spread. Its possible that i legitimately missed after the 3rd or 4th bullet on full auto. I know that my crosshair was right on your chest when i fired the first two with a short hold of the trigger.

Frankie, i wasnt accusing anyone of hiding anything. i was trying to make it clear that i dont suspect anyone of wrongdoing. I know that noone is hiding things, and i want you to know that i know that. Figured it might help if you were at ease about the interaction between us. I am presenting a controversial issue, after all.

I am even more worried about whatever is going on now that i know the hitmesh is 1:1 with the player model. Im not really worried about KDR since i play FA CTF. Im worried about being able to slow someone whos running my teams flag, and im worried about killing people who rush my teams flag area. Basically, i play defense. I usually use burst fire mode, and fire a single burst when my crosshair is over the players chest. My issues have caused a severe lack of tactical capability.

TL:DR im not worried about how i look on the KDR rankings, im worried about how i actually perform in a CTF related setting.

This post has been edited by vareul: 02 February 2014 - 03:53 AM