1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
  2. Become a member of PRTA today, fill out the application form here.
  3. Hey Guest, It'd be great if you'd consider making a small contribution here. All support is greatly appreciated!
  4. Contact! - End of Campaign. Read more here.

Server Issue Server Issue Report - Danesh_italiano

Discussion in 'Player, Admin and Server Reports' started by Danesh_italiano, Sep 9, 2019 at 15:46.

  1. Offline

    Joined:
    Dec 23, 2017
    Messages:
    37
    Likes Received:
    28
    Trophy Points:
    70
    Squad Name:
    Danesh_italiano
    PR:BF2 Name:
    Danesh_italiano
    Server Report

    Report Submitted by: Danesh_italiano

    Server: Project Reality - Modern Conflict

    Nickname: Danesh_italiano
    Profile:


    Date of Issue: Sep 9, 2019
    Time of Issue: 15:00:00

    Description of the issue: Server is running very poorly. Any time you do !fps, it will report lot of bad/high frames! Does not matter if the server is full or have only 20 players (just tested with 20 players on Fallujah and reported 5 bad frames above 80ms).

    You can see the problem on my video but if you join PRTA right now and keep typing !fps, it will report bad/high frame ALL THE TIME! And this problem may be impacting the deviation system?!?!?! (Alon know better about this)

    MY offer for testing with windows server still there. (i know a easy solution to "enable" prta cp/back_end/what ever you call it on my server...)

    FIX IT PL0X!

    Evidence:

    I agree that I have read and understood the PRTA Server Rules :

    ___________________________________________________________________________

    An internal Server Report has been created in the Administrator forums to track and monitor this technical issue.
    Thank you for your report.

    An official statement will be posted if nessarsary within 7 days from the initial response.

    PRTA Administration Team
    Last edited: Sep 9, 2019 at 16:07
    solidfire93, hairyballs and PBAsydney like this.
  2. Offline

    Joined:
    Mar 13, 2017
    Messages:
    10
    Likes Received:
    19
    Trophy Points:
    21
    PR:BF2 Name:
    [R-DEV]AlonTavor
    Linux isn't the issue. Something with the set-up is.
    If you say the forbidden phrase "Windows Server" on these forums even less will be done about it.


    tobi-the-fraggel The "temp" folder used by tracker (serverroot/temp): can you make it tmpfs (64mb should be enough)? From what I've seen from the profiler log (from 2 months ago) there are weird spikes after I/O, even on an almost empty server.
    agus92 likes this.
  3. Online

    Events AREA 94

    Joined:
    May 19, 2014
    Messages:
    2,102
    Likes Received:
    1,046
    Trophy Points:
    226
    Location:
    United States
    PR:BF2 Name:
    Mattytoosack
    Have you tried turning it on and off?
  4. Offline
    PBAsydney PBAsydney Bad hit detection

    Donor

    Joined:
    Dec 8, 2017
    Messages:
    109
    Likes Received:
    76
    Trophy Points:
    115
    Location:
    Iceland
    Discord:
    PBAsydney#1414
    PR:BF2 Name:
    PBAsydney
    I'm getting the same 75-85 every time I do !fps, it's literally unplayable unless I'm using splash damage or networked projectiles.
  5. Offline
    Sloan Sloan Proper planning prevents piss poor performance.

    Technical Dept. Head Sysadmin Lead Coder

    Joined:
    Jul 5, 2012
    Messages:
    13,805
    Likes Received:
    6,453
    Trophy Points:
    276
    PR:BF2 Name:
    Sloan[EEF]
    Server performance is a function of network topology and load as well as the host setup. There have recently been some large-scale attacks on wikipedia which could certainly be causing ongoing issues for various parts of the network, and we know from experience that these can continue for several days. As nothing has changed on the host at the time of reporting it seems most likely that the wikipedia attack is the root cause of any perceived issues.
  6. Offline

    PR:BF2 Administrator 9ª Compañía de Infantería

    Joined:
    Sep 9, 2017
    Messages:
    603
    Likes Received:
    1,765
    Trophy Points:
    151
    Discord:
    Jumbo#6078
    PR:BF2 Name:
    Alfz|Jumbo
    preppies
  7. Offline

    Joined:
    Dec 23, 2017
    Messages:
    37
    Likes Received:
    28
    Trophy Points:
    70
    Squad Name:
    Danesh_italiano
    PR:BF2 Name:
    Danesh_italiano
    [​IMG]

    That is not a network issue. I made those videos 25/08/2019....
    solidfire93, Lorfah and mectus11 like this.
  8. Offline

    Joined:
    Mar 13, 2017
    Messages:
    10
    Likes Received:
    19
    Trophy Points:
    21
    PR:BF2 Name:
    [R-DEV]AlonTavor
    And this is why I don't bother chasing PRTA. I can't do anything when you refuse to admit the problem.

    Network stress will not create frametime spikes and the problem has been going on for at least 6 months.

    I showed you how to use the Python profiler.
    I explained why the "inaccuracy of the python timer" I (unfortunately) mentioned before exists and how its completely irrelevant if a spike is above 66ms.
    In case you still don't trust the Python timer I went on and found the place in memory the server uses to store the current frametime. (sent to Inspect when I talked to him 3 weeks ago about the PRTA lag. float at [0x0108D9F0] + 0x48). This is used for the game simulation.


    Received nothing back.

    BF2 goes to lengths to hide the fact the server is lagging. Its set to allow 10 tick skips without a slowdown. You can run a server on 3.5hz FPS and people will barely notice. Visual latency will be half a second late but hitreg will barely function.

    1.6 will have a "!fps report" function that toggles on/off sending you a message every time server tick took more than 70ms, so its easier to see how often server lags.
    Last edited: Sep 10, 2019 at 12:06
  9. Offline
    PBAsydney PBAsydney Bad hit detection

    Donor

    Joined:
    Dec 8, 2017
    Messages:
    109
    Likes Received:
    76
    Trophy Points:
    115
    Location:
    Iceland
    Discord:
    PBAsydney#1414
    PR:BF2 Name:
    PBAsydney
    A while ago SSG had two PR servers running, one of them based in France running on Linux, the other based in Germany running on Windows. AFAIK they were running on similar hardware. It was random which one of them would seed during the early day but it was always very noticeable once you started playing which server you were on without checking. Their Linux server had the same hitreg and performance issues we are experiencing with PRTA and HOG these days, while their Windows server was performing well. There might be more to this, but I'm convinced there is a clear performance gap between windows and linux based servers. Heck, even the russian server running on windows with double the ping of PRTA has way better hit detection and no noticeable micro-rubberbanding last time I played on it.

    And it's not like I have a Windows bias, if anything I'm biased towards Linux in 99% of cases, but the BF2 servers just seem to run better on Windows.
    Rooky_Ghost and mectus11 like this.
  10. Offline

    PR:BF2 Lead Administrator

    Joined:
    Jan 23, 2013
    Messages:
    826
    Likes Received:
    666
    Trophy Points:
    181
    PR:BF2 Name:
    Filamu
    Okay, so what is causing this, or will this profiler show what causes it?
    What can we do to fix it?
    Rooky_Ghost likes this.
  11. Offline

    Sysadmin IRL Event Attendee

    Joined:
    Apr 12, 2016
    Messages:
    744
    Likes Received:
    893
    Trophy Points:
    181
    PR:BF2 Name:
    Inspect
    During the months this whole talk has been going on, I wouldn't say nothing has changed. There's always small stuff changing, but nothing major that should affect the server has changed afaik.


    You are probably aware that there's a lot of other things that can end up having the same effect aside from OS. How do you know the two SSG servers were running same configs? Similar filesystems? Same networking? They probably were not.

    Micro-rubberbanding is an interesting mention, especially every time I try to ask about how does it specifically occur and does it have to be a full server, people close up. If there would be a way to eg. video record this microstuttering, we could test every single config parameter and setup individiually and record the results. It would be a linear way of troubleshooting and would end up having actual results instead of just "I feel this is better now."
    --- Double Post Merged, Sep 10, 2019 at 13:58, Original Post Date: Sep 10, 2019 at 13:52 ---
    Oh shit, sorry. You can blame me on this one. I totally forgot that you DMd me about this. Had my mind on the IRL event and at the time I didn't see it through. :oops:
    agus92 likes this.
  12. Offline
    PBAsydney PBAsydney Bad hit detection

    Donor

    Joined:
    Dec 8, 2017
    Messages:
    109
    Likes Received:
    76
    Trophy Points:
    115
    Location:
    Iceland
    Discord:
    PBAsydney#1414
    PR:BF2 Name:
    PBAsydney
    I spoke with rana about the servers back then, and as far as I remember both boxes were running an overclocked 4790K with similar specs. Obviously a windows system is not gonna be running the same filesystem and networking as a linux system.

    The micro-rubberbanding I haven't noticed lately, I think it mostly stopped after some server-side patch was applied a few months ago. It was like someone kept jerking the mouse, if i were to aim at a specific point, half a second later the aim would rubberband to a different spot, but that seems to be mostly fixed, and I appreciate that.

    But bullets doing no damage is still very much a thing and is driving me insane. If I am playing as AR I might be able to down a moving target with a quick burst no problem, then 30 seconds later a stationary target tanks 15 bullets and doesn't give a shit. Being on the receiving end is also common, you take automatic fire and you hear all the bullets hit your player model but take absolutely no damage.

    This happens regularly:
  13. Offline

    Joined:
    Mar 13, 2017
    Messages:
    10
    Likes Received:
    19
    Trophy Points:
    21
    PR:BF2 Name:
    [R-DEV]AlonTavor
    That server-side patch increases the update rate from 20hz to 30hz. The expected outcome is less latency for everyone: less getting shot around corner, less delay for your shots to have an effect.
    Any other sideeffects are unknown. Considering other game engine standards (Source,Quake(CoD),Unreal,Overwatch) hitreg should not be affected by adjusting the update rate.

    PBAsydney Do you have a full video showing the last time you changed away from Weapon index 4?
    Deviation is not synchronized. Changing weapons at the same time you go prone/jump/do something that increases deviation might register that deviation penalty on the other weapon server-side. I bet this might happen more often if server skipping ticks
  14. Offline

    Sysadmin IRL Event Attendee

    Joined:
    Apr 12, 2016
    Messages:
    744
    Likes Received:
    893
    Trophy Points:
    181
    PR:BF2 Name:
    Inspect
    Is there a way to reproduce this so that one can join the server, test hitreg instantly, change parameters, reboot the server and join again to test how it works now? Running server for two months after every new parameter change and at the end of it making a player questionnaire about hitreg and ruling out possible biases is not really a viable way with this.
    Rooky_Ghost likes this.
  15. Offline

    Joined:
    Mar 13, 2017
    Messages:
    10
    Likes Received:
    19
    Trophy Points:
    21
    PR:BF2 Name:
    [R-DEV]AlonTavor
    Deal with the framespikes first, which are easy to test, then see if it helped.
  16. Offline
    PBAsydney PBAsydney Bad hit detection

    Donor

    Joined:
    Dec 8, 2017
    Messages:
    109
    Likes Received:
    76
    Trophy Points:
    115
    Location:
    Iceland
    Discord:
    PBAsydney#1414
    PR:BF2 Name:
    PBAsydney
    I don't unfortunately, I'll try to get it recorded.

    In theory, weapons having minimum deviation clientside but a lot of deviation server side for whatever reason could actually explain a lot of hitreg problems, I'm seeing everything connect but from the server's point of view I'm just spraying bullets everywhere. Could be caused by "3-Z-4" too quickly causing desync between client and server deviation? But that doesn't explain why stationary enemies are very often immune to vehicle coax, or why weapons without a deployed mode are frequently not doing damage regardless of visual hits client side.

    Not really. Bolt action / sniper rifles are the most noticeable. Sometimes you can hit the same stationary target with 3 shots in a row and do no damage. But the hit detection problems seemingly come in random spikes, sometimes you have issues, sometimes it's fine.
  17. Offline

    Joined:
    Mar 13, 2017
    Messages:
    10
    Likes Received:
    19
    Trophy Points:
    21
    PR:BF2 Name:
    [R-DEV]AlonTavor
    That's my thoughts. Framespikes might make it happen more often. You apply the deviation penalty to index 3, server applied it to index 4.

    No idea.
    Last edited: Sep 10, 2019 at 18:41

Share This Page