Hello,

I am having some stability problems with the following LINUX system
and would really appreciate some advice.
Note: our latest stress-tests seem to show that things become (more) stable
if instead of 64M Cache Non-ECC we use only 16M (1 SIMM instead of 4 -
things become unstable with 2, 3, or 4 of them and look stable just with 1
SIMM only)

System Config:

Tyan 1832DL with
   Dual 400MHz Intel Pentium II processors
   256M of Tyan recommended memory (pricey stuff)

Dual channel SmartRAID-IV 3334UW controller connect to:
        64M Cache Non-ECC (cheaper standard memory)
     External disk storage with 6 4.3 Gig Western Digital 4360's External
storage unit attached to single channel of 3334UW
     Channel 1  is connected to  the 6, External disks divided into two 12
Gig stripe sets.
     Channel 0 is left open for future expansion.

Adaptec AHA-2940 UW  connected to:
    2 internal 4.3 Gig Western Digital 4360's on scsi II connector (System
disk, /usr and /var)
    Sony Tape Unit & CDRom  on scsi I connector

Basic Trident PCI, VGA  card
3Com 3c509 NIC (I couldn't keep it running for more than an hour with a
3c905)

RedHat Version 5.2 (2.0.35 Kernel)

I put the system together by loading the LINUX system on the two
internal disks attached to the Adpatec controller. I configured RedHat
Linux to use the eata-dma driver to recognise the DPT RAID card. I
rebuilt the kernal to use both cpu's.  I created two striped disk sets
of three drives each in the external enclosure using the the DOS based
Storage Manager.  I created 2, 12 gig ext2 file systems on the RAID
units.

It all works at least for a few days at a time.

Symptoms:
- When the system crashes, It trashes the file systems attached to the
3334UW RAID controller and requires a tape restore.

- About once a week the system hangs and I see no messages. Instantly
freezes the screen.


Have I put together a marginally stable group of hardware here or is
there some secret I have missed about all the configuration issues of
my selected hardware components?  I would love some insight into this
since we are about to bite the bullet and switch over to a Solaris
operating system in hopes of stabilizing on our hardware.  We love
LINUX and do not want to switch but I am at my wits end.

Thanks!

Otis

Reply via email to