hello. my 1on1 server crashes every third day in average. I use Linux and can only see parts of the crash report when opening the screen, this is the part I can get of it:

Code: Select all

^7======
^7server EDICT 2:
^7absmin         '-1679.2599   520.5430   246.5503'
^7absmax         '-1677.2599   522.5430   248.5503'
^7movetype           8.0000
^7origin         '-1678.2599   521.5430   247.5503'
^7oldorigin      '-1112.0000  2032.0000    27.0000'
^7velocity       '   -2.3346   399.9915    -1.4259'
^7angles         '    0.0000    90.2252     0.0000'
^7classname      player
^7model
^7think          SUB_Null()
^7health          6666.0000
^7frags           -666.0000
^7weaponmodel
^7chain          entity 1
^7view_ofs       '    0.0000     0.0000    35.0000'
^7button2            1.0000
^7impulse            7.0000
^7v_angle        '    0.1758    90.2252     0.0000'
^7netname        ^0II^1`^7Luna.
^7colormap           2.0000
^7team               5.0000
^7watertype         -1.0000
^7speed            400.0000
^7playermodel    models/player/skadi.zym
^7playerskin     0
^7killcount       -666.0000
^7welcomemessage_time    11.6300
^7welcomemessage_time2    99.9100
^7version          150.0000
^7jointime           3.6300
^7clientcolors      68.0000
^7movement       '  400.0000     0.0000     0.0000'
^7cursor_trace_start '-1678.3013   523.0263   282.2214'
^7cursor_trace_endpos '-1688.3208  3071.9688   274.4013'
^7ping              60.0027
^7Host_Error: server: Program error
^7QuakeC crash report for server:
^7STORE_F    3500(IMMEDIATE)1000000.0000 13(?)
^7STORE_F    1(?)                 16(?)
^7CALL5      3469(Spawn_ClassifyPoints)Spawn_ClassifyPoints()
^7STORE_ENT  1(?)                 3495(spot)
^7IF         3495(spot)entity 0   branch 8
^7IFNOT      3494(anypoint)    1.0000 branch 4
^7STORE_F    3501(IMMEDIATE)PutClientInServer: no start points on level 4(?)
^7CALL1      206(error)error()
^7gamec/cl_client.c : SelectSpawnPoint : statement 86
^7gamec/cl_client.c : SelectSpawnPoint : statement 89
^7gamec/cl_client.c : PutClientInServer : statement 18
^7gamec/cl_client.c : PlayerPreThink : statement 260
^7Client "cYan" dropped
^7^4cYan^4 disconnected
^7Client "^0II^1`^7Luna." dropped
^7^4^0II^1`^7Luna.^4 disconnected
^7Quake Error: Host_Error: server: Program error


any ideas ?

btw, me and also other players noticed strange lag. some people say it's in regular intervall, but I cannot confirm that. for me it happens approx. once in a map (15 minutes) for about 10 seconds. it simply starts lagging, everything lags, shots, movement, etc. ... I experienced that also in the phase when LH tried to track the "overflow" bug ... is is back ? I don't hope so.

it could of course be also caused by the server it self (other processes), but I can't think of any process making the game-process underpowered for more than 0,5 seconds...

EDIT: I just got to know that this error occured directly after trying to switch the map via vcall ... it's because many people don't know how to change maps and start messing around with things they have no clue about

GreEn`mArine wrote:
EDIT: I just got to know that this error occured directly after trying to switch the map via vcall ... it's because many people don't know how to change maps and start messing around with things they have no clue about

It wasn't me, not this time !!!

(was it me????)

no it wasn't you. please don't spam xD but let the "pro"s have a look at this ^^

Hmmm this line

Code: Select all

^7STORE_F    3501(IMMEDIATE)PutClientInServer: no start points on level 4(?)

Seems to say that there are no spawnpoints in this map..?

Does the server crash compleatly (app close) or does it just "stop"?

I was in "bash" (shell) when I opened the screen for the server ("screen -r")

strange. as i remeber if the server cant find the map, or the map is broken it drops ya to the console, in the case of a dedicated im not sure.. is it possible taht the server software/hardwaare is faulty? (maybe a broken memory chip?)

The server doesn't check if a map is valid when users use "vcall map/chmap". I've just fixed that, but i don't see what could be wrong with your server... "chmap bogus_map" doesn't crash the server in my test and i can't think of a reason why there are no spawnpoints found.

Can you please start the server with "-condebug" and mail me the log the next time it crashes?

I believe this has been addressed in 1.5.1
unfortunatly, a cvs engine will not work with 1.5, so if you want to patch the 1.5 code, you would need to make the following change:

http://cvs.icculus.org/cvs/twilight/dar ... ortby=date

tZork wrote:strange. as i remeber if the server cant find the map, or the map is broken it drops ya to the console, in the case of a dedicated im not sure.. is it possible taht the server software/hardwaare is faulty? (maybe a broken memory chip?)

A hardware fault would probably have caused a segfault, it's very unlikely to cause the qc vm to crash with a detailed error message.
And all system processes are running stable on that machine, so i would say it's definitely a qc/engine thing.

@Willis: do get this right? It's caused by clients that timeout on connect? Did your servers give the same error when they crashed?

ah, you're right KadaverJack, that was the engine related problem, this is cearly something with qc (no, I didn't get a qc crash, just an engine crash where logging would end with a player disconnect).

well, I started the server using "- condebug +exec server.cfg" after the binary ... but where should this log file be created ? and when ? I can't see any logfile directly after starting up

it should be -condebug
in which case it will create a file called "qconsole.log" in your data directory

Willis wrote:in which case it will create a file called "qconsole.log" in your data directory

which is ~/.nexuiz/data on Linux

KadaverJack wrote:

Willis wrote:in which case it will create a file called "qconsole.log" in your data directory

which is ~/.nexuiz/data on Linux

ah, I didn't know it was in a hidden directory (which is infact a pretty stupid idea to have it there instead of having it in the same nex directory where the binaries are)

GreEn`mArine wrote:ah, I didn't know it was in a hidden directory (which is infact a pretty stupid idea to have it there instead of having it in the same nex directory where the binaries are)

No, that's actually pretty clever as the user starting the Nexuiz binary doesn't necessarily have the right to write into the same nex directory where the binaries are.

GreEn`mArine wrote:

KadaverJack wrote:

Willis wrote:in which case it will create a file called "qconsole.log" in your data directory

which is ~/.nexuiz/data on Linux

ah, I didn't know it was in a hidden directory (which is infact a pretty stupid idea to have it there instead of having it in the same nex directory where the binaries are)

As Savage said, it clever. Itas also standard procedure in most *nix eviroments, taht programs create dot-directories in the running users home directory for their settings that are user spesific.

The *nix-enviroments are true multi-user enviroments, and users generally cannot write to anywhere besides their own file structure (home dir)

The reason for hiding the directory (done by prefixing the dir name with ".", same goes for files) is that elsewise you'd end up with a pretty messy home dir. Most of the people i know tend to use ls -la (list and all) by default when browsing, this to display also hidden files.

Hehe its a idea hard to get for windows users. Actually i like programs that do it this way on windows too. To bad quake/darkplaces and most other programs do not . Some games like deus ex 2 did place savegames into your 'own files' directory. Thats cool because that why several users can play that game and to do have to worry about others settings/savegames.