got to post this because I'm running out of options. And patience
:). I'm hosting some half-life and qw servers on this linux box, but not
log ago we upgraded from a p5/166/66fsb to k6-2/450/100. Because money are
a problem the box has to do briging with 4 network adapters (2pci/2isa -
realtek 8029 and via 86c926 I think, bnc 10mb/s). now from time to time I
get VERY HIGH loads, computer almost frezees. It wasn't ftp from/to the
server; the games client-server talk takes about 3~4k/player (I used
r_netgraph 1 in hl console). it's the brdging the causes me those
problems. I tried more versions, now works pretty stable with
bridgex-0.30, kernel 2.2.15-pre19. Still got high loads.
here a dmesg copy.
----------------------
Linux version 2.2.15pre19ide ([EMAIL PROTECTED]) (gcc version egcs-2.91.66
19990314/Linux (egcs-1.1.2 release)) #1 Tue Apr 18 21:29:29 EEST 2000
Detected 451020794 Hz processor.
Console: colour VGA+ 80x25
Calibrating delay loop... 897.84 BogoMIPS
Memory: 63592k/65536k available (664k kernel code, 412k reserved, 832k data, 36k init)
Dentry hash table entries: 8192 (order 4, 64k)
Buffer cache hash table entries: 65536 (order 6, 256k)
Page cache hash table entries: 16384 (order 4, 64k)
CPU: L1 I Cache: 32K L1 D Cache: 32K
CPU: AMD AMD-K6(tm) 3D processor stepping 0c
Checking 386/387 coupling... OK, FPU using exception 16 error reporting.
Checking 'hlt' instruction... OK.
POSIX conformance testing by UNIFIX
PCI: PCI BIOS revision 2.10 entry at 0xfb220
PCI: Probing PCI hardware
Linux NET4.0 for Linux 2.2
Based upon Swansea University Computer Society NET3.039
NET4: Unix domain sockets 1.0 for Linux NET4.0.
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP
TCP: Hash tables configured (ehash 65536 bhash 65536)
Initializing RT netlink socket
Starting kswapd v 1.5
Detected PS/2 Mouse Port.
pty: 256 Unix98 ptys configured
Uniform Multi-Platform E-IDE driver Revision: 6.30
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
ALI15X3: IDE controller on PCI bus 00 dev 78
ALI15X3: not 100% native mode: will probe irqs later
ide0: BM-DMA at 0xf800-0xf807, BIOS settings: hda:DMA, hdb:pio
ide1: BM-DMA at 0xf808-0xf80f, BIOS settings: hdc:pio, hdd:pio
hda: FUJITSU MPB3021ATU, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hda: FUJITSU MPB3021ATU, 2062MB w/0kB Cache, CHS=523/128/63, UDMA(33)
NET4: Ethernet Bridge 007 for NET4.0
early initialization of device brg0 is deferred
brg0: network interface for Ethernet Bridge 007/NET4.0
brg0: generated MAC address FE:FD:0B:81:AF:FA
brg0: attached to bridge instance 0
Partition check:
hda: hda1 hda2 hda3 hda4
VFS: Mounted root (ext2 filesystem) readonly.
Freeing unused kernel memory: 36k freed
Adding Swap: 40280k swap-space (priority -1)
ne2k-pci.c:vpre-1.00e 5/27/99 D. Becker/P. Gortmaker
http://cesdis.gsfc.nasa.gov/linux/drivers/ne2k-pci.html
ne2k-pci.c: PCI NE2000 clone 'RealTek RTL-8029' at I/O 0xf000, IRQ 11.
eth0: RealTek RTL-8029 found at 0xf000, IRQ 11, 52:54:05:E2:AA:90.
ne2k-pci.c: PCI NE2000 clone 'RealTek RTL-8029' at I/O 0xf400, IRQ 10.
eth1: RealTek RTL-8029 found at 0xf400, IRQ 10, 00:C0:DF:EF:53:03.
device eth0 entered promiscuous mode
device eth1 entered promiscuous mode
ne.c:v1.10 9/23/94 Donald Becker ([EMAIL PROTECTED])
NE*000 ethercard probe at 0x340: 00 00 21 4a 54 47
eth2: NE2000 found at 0x340, using IRQ 9.
NE*000 ethercard probe at 0x260: 08 00 00 00 05 83
eth3: NE2000 found at 0x260, using IRQ 4.
device eth2 entered promiscuous mode
device eth3 entered promiscuous mode
br: enabling port 1
br: enabling port 2
br: enabling port 3
br: enabling port 4
br: enabling bridging function
---------------------------
I observed that those high loads appear when users copy large
amount of data from one another (ftp or win exlporer copy), not from the
server. that's bridging. my ping with the server goes up to 20-50ms and
the box hardly respond to any commands.
procinfo -D -n1 dump
---------------------------
Linux 2.2.15pre19ide (root@matrix) (gcc egcs-2.91.66) #1 1CPU [matrix]
Memory: Total Used Free Shared Buffers
Cached
Mem: 79848 78308 1540 11156 31724
31840
Swap: 40280 1620 38660
Bootup: Fri Jun 9 02:06:55 2000 Load average: 1.46 0.50 0.27 2/30 3862
user : 0:00:00.00 0.0% page in : 0 disk 1: 0r
4w
nice : 0:00:00.00 0.0% page out: 4
system: 0:00:02.24 96.6% swap in : 0
idle : 0:00:00.08 3.4% swap out: 0
uptime: 23:39:19.75 context : 19
irq 0: 232 timer irq 10: 406 eth1
irq 1: 0 keyboard irq 11: 16 eth0
irq 2: 0 cascade [4] irq 12: 0 PS/2 Mouse
irq 4: 237 NE2000 irq 13: 0 fpu
irq 9: 261 NE2000 irq 14: 4 ide0
-----------------------
96.6%? and no game servers are running. I think someone/eth1 is making a
ftp transfer with isa-eth3/irq4 (this dump).
now there's something else bothering me. ok..150-200-500ms pings in
half-life when people are copying stuff. of course the game is unplayable,
lag's too big. but this was with the old pentium/166. how come a
k6-2/450/100fsb acts the same?!? the system is 3-4times
quicker and even with full network load I think it can handle a 1meg
bandwidth.
really odd. some of the guys who helped me buying the new hardware are
pissed off because we're getting the same results...no progress.
I could use the old p55c/166 just for bridging and the k6 for game
hosting. or someone could give me a hint about traffic limitations in
briging. is there a (new) patch for this high system load? I'm on rh61,
2.2.15-pre19, bridgex 0.3.
please give me some directions. thanks for your time.
/dx
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
