My list of Server bugs

[gravypod]

1. Server waits on pongs
2. Server doesnt remove a client when he disconnects leading to the server holding up on a ping/pong from a null client
   
   not executing server ping for null client yet: 0 / 10000
   [SERVER][PROCESSOR] client setup completed. listening for input
   [SERVER][LOGIN] new client connected. given id: 15: description: TYLER377190
   [SERVER][LOGIN] return code 0
   not executing server ping for null client yet: 5001 / 10000
   not executing server ping for null client yet: 1000 / 10000
   not executing server ping for null client yet: 6001 / 10000
   not executing server ping for null client yet: 2001 / 10000
   not executing server ping for null client yet: 7001 / 10000

[*]Kicks (seem to) require a packet to be acepted by client
[*]you can buy a ship anywhere. It says you can't buy it but then it spawns one for you. Only admins should be able to do this.
[*]Teleport_self_to doesnt work, plops you in the middle of a sector
[*]No ping responce should just kick the client. http://docs.oracle.com/javase/6/docs/api/java/net/Socket.html#close()
[*]collision warping doesnt work
[*]"Server saving error" turns on a shut down, within that time the server saves its data correctly :P
[*]The shutdown command should save everything before exiting, dont use System.exit(0), close connections, threads, save all data, close SQL Server, the works. Dont rely on shutdown hooks, some OSs doent support them
[*]Server commands that dont require the StarMote via terminal
[*]Improve server logging, only log chat, lag messages and warnings. We dont need to see things like "[DELETE][Server(0)] Sendable 1883(kc@275d00ef) Physically DELETING DONE and Notified!". Using .toString() is a heavy operation. Adding a debug option to the server.cfg and defaulting to off would save a lot of CPU time.
[*]Server logs are deleted when they fill up, I dont know how this makes them usefull
[*]Exiting starmote kills the game and exits the window, it should just bring you to the login screen
[*]The "not executing server ping for null client yet: 0 / 10000
java.io.EOFException
at java.io.DataInputStream.readInt(DataInputStream.java:392)
at org.schema.schine.network.server.ServerProcessor.run(SourceFile:777)"
[*]Don't warp out of collision, fix your physics engine to that it doesn't lock up the server
[*]AI ships dont work/spawn on my server... for no reason
[*]Exiting a ship core never puts you next to a ship, always below the the right
[*]"warning CollisionDispatcher.needsCollision: static-static collision!
warning CollisionDispatcher.needsCollision: static-static collision!"
[*]Physics are don't client side sometimes causing problems (Ships arent where they appear)
[*]There should be a central skin registry like minecraft has, skin tied to the account so that serversdoesnt need a copy.
[*]Seems like you wait on SQL Query

1. 
   [SQL] SECTOR QUERY TOOK 77ms types: (1,2); from (18, 24, 6) to (26, 32, 14)

I will keep updating this as time goes on

 

[schema]

I\'m on fixing the server EOF issues. It seems like depending on OS and machine, a last flush is not made before a socket is closed (although I had read that it does). I relied on software timeouts which didn\'t work to well. I added a socket timeout now that hopefully will take care of leftover connections.

The buy ship anywhere bug is left in on purpose so people can buy something if they are stranded. But I can see how that will become a problem with more serious players. I\'ll remove it (or maybe make it a server option)

Thanks for the tip. I will make a \"normal\" shudown without shutdown hooks.

Server commands via terminal and remotely will also come soon.

So will a second \"readable\" log.

I currently can not afford the extra bandwidth for a central registry, but I think about it when we upgradethe servers.

I can\'t really \"fix\" the physics, the only way to safely do a collision would be to cast the object, which pretty much cripples perfromance. Collisions are done in a snapshot way, as it\'s the usual for physics engines. The main reason people are getting stuck is upping the max speed, which puts the snapshots farther apart per update. Depending on the speed of the cpu, less snapshots can be processed by frame. So something can happen called \'tunneling\', which is one of the most difficult things to prevent. The warp out of collision algo has gotten a lot better, but it\'s of course not perfect, as I have to check the next set of objects that cause problems after I updated it last week.



The problem with AI ships have been reported, but I still need a way to reproduce the issue.



The sql query in question is a galaxy-map query, but those are done in a thread (or if not yet, will be), so it won\'t lock up anything.

 

[gravypod]

I would love to help debug the AI ship bugs. Just pm me and we can work it out together. Also, why do you need to cast the object, as long as you arent storing it in a raw map/collection there should be any casting needed.

 

[TerrorSoul]

\"Server commands via terminal and remotely will also come soon.\", That is the best news to hear considering myself and probably alot of other server owners have tried and failed and indeed come to the conclusion that at the current momment its impossible to run commands remotley from servers :/, the quicker that functionality is added the better

 

[schema]

cast what object? I\'m doing a lot of stuff with generics, but in a codebase that large, not casting some objects is not really possible. I\'m trying to avoid it as best as possible.

 

[TerrorSoul]

A small question but nonetheless, How far into development until Remote Commands/ Ability to Access Server Console online and execute commands will be added?



It\'s got to the point where I\'ve actually considered getting a VPS and running StarMote on it along with some sort of bot in order to somewhat automate my server.

 

[Racr]

People have already figured out how to run the server on a linux vps and get it to automatically restart itself. Even though there are no console commands yet, I\'ve no idea how they do it though.

 

[gravypod]

#!/bin/bash
cd \"$(dirname \"$0\")\"
while true
do
java -jar StarMade.jar -server
done

exit $?

 

[TerrorSoul]

You can only restart a server, everything else is impossible me and a few others have tried our bests and its impossible at the moment. I spoke to 8bit the owner of SMPROHOSTING the other day and he had the same issues.

 

[gravypod]

[WARNING][SERVER] Exception: synchronized flush took 400 ms, size: 1597 bytes