Server hang AGAIN (aiee:scheduling in interrupt)

Dear Squid-Admin,

        Subject from the squid-admins' suggestion, we upgraded our linux box from
2.0.30 to 2.0.33. (Current configure : Pentium-133, aha2940uw, 3c905, 1G
ide-mode4 hd for system, 4G U-fastwide disk for cache data, 128Mram,
squid-1.1.20)

        However, the box is hang again at mid-night (0:00am) with the same way
"aiee:scheduling in interrupt". During that time, the machine is running
"cache-stats-1.4" statistics package. Normally, the free memory (free -m)
only left 1 Mbyte during daily operation. Hence, I guess, the problem is
coming from "not enough memory" used for "squid proxy", "cache-stats-1.4"
as well as OS itself. In addition, the IRQs of disk-controller is irq 11
and 3c905 is irq 15. We tried to push 3c905 irq other than 15 or 9, but fail.

        As we have other squid-linux box with OS equal to 2.0.29, squid 1.1.20,
physical memory = 180M, disk-controller is aha2940uw (irq11), 5 x 1G scsi-2
HD for system and cache, 3c905 Full-duplex (irq10), CPU P5-90Mhz, it run
stable.

Therefore:

(1) Should I down grade the OS from 2.0.33 to 2.0.29 ?
(2) Should I use NOVM version ?
(3) Should I change OS from Linux to freebsd ?
(4) Should I remove the IDE hd and move linux OS to U-Fast Wide disk ?

        Could anyone tell me how to overcome the problem ? Sorry for any
inconvenience caused.

Thanks,

Lopy
Hong Kong Baptist University

Addition information (Capture from : dmesg)

Console: 16 point font, 400 scans
Console: colour VGA+ 80x25, 1 virtual console (max 63)
pcibios_init : BIOS32 Service Directory structure at 0x000fe100
pcibios_init : BIOS32 Service Directory entry at 0xfa96d
pcibios_init : PCI BIOS revision 2.10 entry at 0xfa310
Probing PCI hardware.
Calibrating delay loop.. ok - 53.04 BogoMIPS
Memory: 127884k/131072k available (696k kernel code, 384k reserved, 2108k
data)
Swansea University Computer Society NET3.035 for Linux 2.0
NET3: Unix domain sockets 0.13 for Linux NET3.035.
Swansea University Computer Society TCP/IP for NET3.034
IP Protocols: ICMP, UDP, TCP
Checking 386/387 coupling... Ok, fpu using exception 16 error reporting.
Checking 'hlt' instruction... Ok.
Intel Pentium with F0 0F bug - workaround enabled.
alias mapping IDT readonly ... ... done
Linux version 2.0.33 (root@mercury) (gcc version 2.7.2.3) #1 Tue Feb 17
13:54:31 CST 1998
Starting kswapd v 1.4.2.2
Serial driver version 4.13 with no serial options enabled
tty00 at 0x03f8 (irq = 4) is a 16550A
tty01 at 0x02f8 (irq = 3) is a 16550A
ide: i82371 PIIX (Triton) on PCI bus 0 function 57
    ide0: BM-DMA at 0xafa0-0xafa7
hda: WDC AC21200H, 1222MB w/128kB Cache, CHS=621/64/63, DMA
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
Floppy drive(s): fd0 is 1.44M
FDC 0 is a National Semiconductor PC87306
aic7xxx: <Adaptec AHA-294X Ultra SCSI host adapter> at PCI 19
aic7xxx: BIOS enabled, IO Port 0xf800, IO Mem 0xff001000, IRQ 11, Revision B
aic7xxx: Wide Channel, SCSI ID 7, 16/16 SCBs, QFull 16, QMask 0x1f
scsi0 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 4.1.1/3.2.1
scsi : 1 host.
scsi0: Scanning channel A for devices.
  Vendor: IBM Model: DCAS-34330W Rev: S65A
  Type: Direct-Access ANSI SCSI revision: 02
Detected scsi disk sda at scsi0, channel 0, id 0, lun 0
scsi : detected 1 SCSI disk total.
SCSI device sda: hdwr sector= 512 bytes. Sectors= 8467200 [4134 MB] [4.1 GB]
eth0: 3Com 3c905 Boomerang 100baseTx at 0xfcc0, 00:60:97:19:92:79, IRQ 15
  8K word-wide RAM 3:5 Rx:Tx split, MII interface.
eth0: MII transceiver found at address 24.
  Enabling bus-master transmits and whole-frame receives.
eth0: Overriding PCI latency timer (CFLT) setting of 16, new value is 248.
3c59x.c:v0.46C 10/14/97 Donald Becker
http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html
Partition check:
 sda: sda1
 hda: hda1 hda2
VFS: Mounted root (ext2 filesystem) readonly.
Adding Swap: 124988k swap-space (priority -1)
Received on Wed Feb 18 1998 - 20:58:30 MST