On Tue, Feb 10, 2009 at 12:41 PM, Philipp Reddigau <[email protected]> wrote: > We are using 24 Servers with Q9550 4gb ram 1066mhz 2xsata 250gb seagate. > 19 Servers with Q6600 4 GB 800 Mhz ... > Some x3550 > and some amd's. > So more 43 servers have a hardware error? > i don't think so :> it must be a kernel failure but no idea wich module > is broken. > > The AMD's have no failures. >
>>> ---------------------------------------------- >>> CRASH: Mon Feb 9 18:04:48 UTC 2009 >>> Start Line: ./srcds_i486 -game dod +maxplayers 32 +sv_region 3 +sv_lan 0 >>> +map dod_doors_sb_1 +ip 85.131.176.145 -port 28080 -tickrate 100 -debug >>> +fps_max 300 -notv -noipx -timeout 5 +sv_stats 0 >>> End of Source crash report >>> ---------------------------------------------- >>> Their i cant see any reason for a crash *sad*. >>> I have read the kernel log: >>> @Feb 9 18:04:48 babe1781 kernel: srcds_i486[13824] general protection >>> ip:b6256661 sp:bfaf0a4c error:0 in vphysics_i486.so[b61d4000+2da000] I wonder if the code in vphysics_i486.so is either using CPU instructions that are not valid with your Intel based servers (BTW are you running an i386/i686 kernel or an amd64 kernel on them?), or on IA32 / EM64T processors (rather than AMD amd64 processors) there is either an invalid memory access / execution. Perhaps broken logic* on selecting which CPU instruction set to execute (i.e. Extended 3DNow! instruction would fail on an Intel processor) in the vphysics (and possibly other Source ds modules as well). * Or the instruction set test is faulty / has a race condition on multi-code processors. All speculation on my part. _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
