StarMade v0.199.654 - Exploit fixes

    AndyP

    Customer Experience Manager
    Joined
    Aug 15, 2013
    Messages
    1,174
    Reaction score
    254
    I could just screen share my screen to ya and you'd see the evidence for yourself, usually, i'd have to log to main menu and 'clear client cache', which temporarily fixes this and refreshes the de-sync from before you left to make it work normally, until it may de-sync AGAIN and repeat the process.
    Screen sharing does usually not help, we need to make it happen on our systems.
    Either computer directly or on the testserver, as we can monitor the environment and java processes in more detail there to find the exact cause.

    I've found that with large semi-occluded objects on rails the "rail freeze" desync happens "somewhat often". That translates to "a matter of time each session" on one particular build. I'll horse around on my current build over the weekend and see if I can make trigger conditions more reliable than that.

    for now i'll leave a general text description of what I noticed.
    -a "door frame opening" in the main entity covered by force-field blocks
    -4 docked "door panels" that rotate 45 degrees to iris open/closed over the frame
    -the docked "panels" are 1/4 slabs, the "pocket" in the "frame" is 1/2 or 3/4 slabs
    -door will operate normally untile freese occurs(possibly by re-loading occluded chunks during build?)
    -once freese occurs all rails on entity stop motion, although docking beams and pickup points work normally. additionally the large force-field block door will gain 2 square patches that stop cascading( rest of force-field blocks goes tangible/intangible as normal activating touching blocks)

    it might be a collision-check error, but since ALL rails freese activity i'm not too sure. have not checked if it's server-wide, sector wide, or entity chain limited for rail failurw. I'll try to remember to make a couple floating rail-clocks around to verify that.
    Thank you for the help, but general descriptions do not really lead to the target here.
    And especially if doors/force fields act weird like they were single blocks and not in groups, or in general if rails do not move anymore on more than one entity, it is usually a thread-lockup on the server side, and the isolated door toggle is one of the many symptoms of this.
    Other symptoms are for example:
    Players can still play and act normal, but newly connected players stay stuck in the roaming cam screen and can not spawn.
    Or you spawn a core, and get only the target marker, but never the physical core.

    - Andy
     

    DrTarDIS

    Eldrich Timelord
    Joined
    Jan 16, 2014
    Messages
    1,018
    Reaction score
    277
    Screen sharing does usually not help, we need to make it happen on our systems.
    Either computer directly or on the testserver, as we can monitor the environment and java processes in more detail there to find the exact cause.



    Thank you for the help, but general descriptions do not really lead to the target here.
    And especially if doors/force fields act weird like they were single blocks and not in groups, or in general if rails do not move anymore on more than one entity, it is usually a thread-lockup on the server side, and the isolated door toggle is one of the many symptoms of this.
    Other symptoms are for example:
    Players can still play and act normal, but newly connected players stay stuck in the roaming cam screen and can not spawn.
    Or you spawn a core, and get only the target marker, but never the physical core.

    - Andy
    Totally get that "general isn'ttoo helpful" It's just all I could add on that particular computer. Is there a thread-watcher function on the GUI I can bring up with something like F1+I? Or some other convienient debug I can watch once the lock-up happens?
     

    JumpSuit

    Lost-Legacy Director
    Joined
    Feb 5, 2015
    Messages
    344
    Reaction score
    93
    Screen sharing does usually not help, we need to make it happen on our systems.
    Either computer directly or on the testserver, as we can monitor the environment and java processes in more detail there to find the exact cause.



    Thank you for the help, but general descriptions do not really lead to the target here.
    And especially if doors/force fields act weird like they were single blocks and not in groups, or in general if rails do not move anymore on more than one entity, it is usually a thread-lockup on the server side, and the isolated door toggle is one of the many symptoms of this.
    Other symptoms are for example:
    Players can still play and act normal, but newly connected players stay stuck in the roaming cam screen and can not spawn.
    Or you spawn a core, and get only the target marker, but never the physical core.

    - Andy
    I play on Freaks R Us, not the Test Server, I'd need access to the Test Server to even see if I can replicate my client's errors, I'll post screenshots to show the de-syncs:
    Such as Power Regen acting like 1 e/sec when it literally isn't:
    upload_2017-9-2_15-11-48.png
    Or Shield cap Acting like it is 220:
    upload_2017-9-2_15-12-42.png
    Or better yet, my weapons not showing my modules linked to the comps. :
    (merging image)
    Is this better proof?
    [doublepost=1504383378,1504383200][/doublepost] upload_2017-9-2_15-16-9.png
     
    • Like
    Reactions: Agame3
    Joined
    Mar 18, 2014
    Messages
    292
    Reaction score
    153
    Everyone has access to the test server (at play.star-made.org:4242, usually running the latest dev build).
    In software development, reproduction steps are typically requested not to prove that a problem happened, but in order to be able to execute the problem repeatedly in a controlled environment while tracing/debugging the code or examining other details in order to isolate a specific cause.
     

    JumpSuit

    Lost-Legacy Director
    Joined
    Feb 5, 2015
    Messages
    344
    Reaction score
    93
    Everyone has access to the test server (at play.star-made.org:4242, usually running the latest dev build).
    In software development, reproduction steps are typically requested not to prove that a problem happened, but in order to be able to execute the problem repeatedly in a controlled environment while tracing/debugging the code or examining other details in order to isolate a specific cause.
    I got many bug reports on phab waiting responses logs steps and all, and no staff response.
    EDIT: I don't usually run DEV Builds.
     
    Last edited:

    StormWing0

    Leads the Storm
    Joined
    Jun 26, 2015
    Messages
    2,126
    Reaction score
    315
    I got many bug reports on phab waiting responses logs steps and all, and no staff response.
    EDIT: I don't usually run DEV Builds.
    The current dev build is the same as the current live build at the moment but generally most of us have a separate path for the dev build.
     

    JumpSuit

    Lost-Legacy Director
    Joined
    Feb 5, 2015
    Messages
    344
    Reaction score
    93
    The current dev build is the same as the current live build at the moment but generally most of us have a separate path for the dev build.
    Then what is the point of me switching to dev build atm when they the same build as release.
     
    Joined
    Mar 18, 2014
    Messages
    292
    Reaction score
    153
    It means that you nay already be running the same build as the test server (for the time being), and could thus connect without installing the dev build.
     

    JumpSuit

    Lost-Legacy Director
    Joined
    Feb 5, 2015
    Messages
    344
    Reaction score
    93
    It means that you nay already be running the same build as the test server (for the time being), and could thus connect without installing the dev build.
    In theory,:giggle:.